Aug 2025
Tesla Battery Sponsorship
Tesla
Team selected for its high-voltage FSAE battery proposal.
Daniel Ortiz Valencia
Portfolio
Overview
I founded Colorado's first Formula SAE Electric Vehicle program and designed a 324V, 8.64kWh battery system that earned a Tesla sponsorship. Across four competition vehicles, I've designed custom PCBs (BMS, safety interlocks, telemetry), written CAN test frameworks for fault validation, and architected full vehicle harnesses integrating 30+ sensors.
I scaled RAM Racing from 30 to 60+ members, raised $400K, and shipped four competition vehicles, building a team culture around precision, ownership, and real engineering accountability.
4Competition vehicles engineered
60+Members led at RAM Racing
$400KFunding raised
Projects
01Interlatent
Robot teleoperation, hardware integration & agent validation
RoboticsVR / IKQUICAgent Evaluation
Interlatent turns robot demonstrations into training-ready datasets. I worked on teleoperation and agent validation across NORI, Dimensional OS, I2RT YAM, SO-101, and Axol.
Teleoperation Architecture & Hardware Integration
WebXR, inverse kinematics, QUIC & robot hardware
Background
- Needed a shared teleoperation and recording workflow across robots with different joint units, kinematics, cameras, and onboard computers.
Technical Approach
- Integrated CAN, cameras, and host hardware on Raspberry Pi 5 and Jetson; validated clutch mapping, IK, and independent arm control in MuJoCo before live testing.
- Traced WebXR poses through browser IK and QUIC delivery to the robot's command arbiter, freshness checks, joint limits, and motion clamps.
- Evaluated command and camera timing, plus the separate gRPC recording path that spools acknowledged data into LeRobot episodes.
Results
- Helped expand control from one robot to a five-robot fleet through shared interfaces and hardware-specific configuration.
- Identified capture and encoding bottlenecks and documented changes to improve control timing and recording turnaround.
Agent Validation & Reliability Testing
Visual grounding, annotation consistency & workflow reliability
Background
- Dataset agents needed to describe robot recordings and annotate task progress accurately before their outputs could guide training-data selection.
Technical Approach
- Built an eight-area test matrix using cloth-folding and cable-untangling datasets, with explicit prompts, pass criteria, and recorded observations.
- Checked claims against camera frames and state tracks; repeated phase-segmentation and outcome prompts to measure consistency.
- Tracked annotation versions, cancellation, cost, and unintended writes alongside answer quality.
Results
- Exposed ungrounded visual claims, inconsistent phase boundaries, and background writes after cancellation.
- Delivered a reproducible failure log and prioritized fixes for visual evidence, write isolation, and cancellation; confirmed stable outcome labels across three repeated checks.
02Venus EV
Colorado's flagship FSAE Electric car, full HV accumulator & safety stack
HV Accumulator324V BatteryBMSKiCadSystem Integration
As Electrical System Officer for the 2026 car, I managed the LV and HV subteams while personally focusing on energy experimentation to deliver the program's first competition-bound battery pack, balancing the technical demands against the challenge of building it as cheaply as possible so the other subteams had the budget to build the rest of the car.
Battery Pack Architecture & Cell Selection
90s5p · 8.64kWh · 324V nominal, designed cell-level from scratch
Background
- First EV program with no existing battery architecture, required full cell-level design rather than purchasing pre-made modules.
- Budget constraints made in-house cell assembly necessary, saving ~$7,000 but requiring the electrical team to own all pack design decisions.
- Needed to size for FSAE endurance (22km) with enough margin to account for simulation uncertainty.
Technical Approach
- Evaluated five cell chemistries (P30B, P50B, P28A, P60B, M35A); selected Molicel P50B for its energy density (5000mAh, 18Wh) at acceptable weight (71g) and internal resistance (17mΩ).
- Simulated full endurance in OpenLap (2023 Michigan layout), 350Wh/lap, requiring 7.7kWh minimum; applied a 12% margin from a prior ICE vehicle where actual exceeded simulation by 7%.
- Configured as 90s5p (8.64kWh, 324V nominal) in 16 segments of three modules each, enabling symmetric segment removal for future weight optimization.
Results
- Delivered complete pack architecture meeting FSAE energy requirements with 12% validated margin.
- Cell thermal modeling confirmed sub-40°C operation through full endurance, eliminating active cooling complexity.
- Modular design enabled future weight reduction without redesign, ordered 20% extra cells for screening and parallel-group matching.
Battery Pack High-Voltage Integration
Precharge, discharge & isolation monitoring, full KiCad HV architecture
Background
- HV system must precharge inverter bus capacitance before contactor closure, discharge stored energy on shutdown, and continuously monitor isolation failure, any fault is safety-critical.
- FSAE rules require precharge to 95% bus voltage before main contactor closure, discharge to under 60V within a set time, and IMD response within defined latency for isolation loss.
Technical Approach
- Sized 1/0 AWG EXRAD-HVX1V8X cabling from thermal modeling (sub-40°C continuous, margin to 90°C at 500A) and selected Littelfuse protection from datasheet I²t curves.
- Designed precharge (620Ω, 0.88mF bus, 95% in under 10s) and discharge (2.2kΩ, NC relay auto bleed-down) circuits within resistor power ratings.
- Created a full KiCad schematic defining AIR control, IMD integration, TSAL logic, energy metering, and connector interfaces (Amphenol, LEMO, Hirose).
Results
- Delivered complete HV architecture with selections justified by operating conditions and datasheet specs.
- Produced operational documentation covering startup, shutdown, and fault injection sequences.
- System prepared for bench validation of precharge timing and IMD fault response before integration.
Fuse Links
328A short-circuit characterization, benchtop test rig at 200Hz+
Background
- Individual cell fusing must interrupt 328A theoretical short-circuit current to prevent thermal runaway, FSAE requires test data at four current levels (Level 4: 0.05 to 0.15s at highest current).
- Nickel-plated fuse links with reduced cross-section need characterization to validate melt time against the thermal model (t_melt = M·ΔH / (P_gen - P_loss)).
Technical Approach
- Built a benchtop rig using a stick welder (500A short-circuit) with a current sensor feeding Simulink for real-time acquisition and automatic blow-time detection.
- Defined a test matrix across four current levels; a rapid fuse-swap fixture enables iterating geometry to correlate measured blow times with analytical calculations.
Results
- Established a repeatable characterization process with >200Hz sample rate capturing blow transients.
- Test data package prepared for SAE technical inspection, proving the fuse clears fault current within required windows.
Weld & Busbar Quality Validation
~600 in-house spot welds, destructive peel DOE, non-destructive resistance screening, BMS validation
Background
- The nickel fusible link doubles as each cell's fuse, so tab-to-busbar weld quality directly sets a safety-critical parameter; the 90s5p pack required ~450 cells and ~600 spot welds executed in-house with no vendor-qualified process to inherit.
- Needed a measurable, repeatable accept/reject criterion for a hand-operated spot welder before scaling the process across the full pack.
Technical Approach
- Ran a design of experiments across pre-pulse (BI), weld time (WT), weld current (WI), electrode pressure, and pulse count; graded pass/fail by destructive peel test against a failure-mode criterion rather than a force threshold, a passing joint had to tear parent tab material rather than release at the weld interface.
- Low weld time produced cold, low-visual welds and excessive current (WI 15) deformed and holed the cell can, both rejected as under- and over-weld failure modes; the accepted setpoint (WI 10 / WT 4, two pulses) consistently pulled tab material off the cell.
- Followed with non-destructive resistance screening (voltage-drop method) across 8 busbar assemblies × 5 fusible links each, then revised link geometry to reduce nickel path resistance.
- Validated the process in service through the BMS, logging 6 instrumented cells through a full discharge cycle to check for an outlier joint.
Results
- Fusible-link resistance held 0.29 to 0.63mΩ across all 40 screened links, with no outliers flagged for rework.
- Under discharge, cell voltage spread peaked at 28mV (CV 0.32%) across the instrumented group, confirming pack-level uniformity.
- At the pack's ~20A/cell discharge current, that resistance spread alone predicts only ~6.6mV of cell-to-cell variation, well under the 28mV observed, weld/link resistance is not the dominant driver of pack imbalance.
Custom Charging Architecture
6.6kW off-vehicle charger interface with full interlock integration
Background
- FSAE rules require accumulator removal for charging; the system must safely interface an external 6.6kW charger with HV isolation, interlocks, and voltage monitoring via banana jacks.
Technical Approach
- Created a KiCad schematic defining the charger-to-pack interface: EV200A1ANA contactor, Amphenol UPCR012AL51 and TE AMP+HVP800 (50mm²) quick-disconnects, LV25-P voltage transducer with resistor divider for isolated HV sense, and HVIL continuity through a LEMO connector to shutdown.
Results
- Delivered complete charging documentation with connector selections, fuse ratings (30A charger input, 5A energy meter), and wire gauges (8 AWG feed, 84mm² bus bars) enabling off-vehicle charging with full interlock integration.
Custom Battery Management System (in validation)
Distributed 16-slave architecture, 90-wire harness reduced to a 6-wire digital backbone
Background
- Orion thermistor expansion required a 90-wire harness inside a 378V accumulator, creating assembly risk and service complexity.
- Wiring bulk and connector density were the highest-probability failure modes during integration and maintenance.
Technical Approach
- Designed a distributed architecture with 16 slave boards, 1 master, 16-bit ADCs over SPI, isolated GLV-to-tractive domains, and CAN communication to the BMS.
- Replaced parallel analog wiring with a 6-wire digital backbone, local temperature acquisition with centralized data aggregation.
Results
- Currently in the process of validation.
03Rubicon EV
First CSU EV car, full vehicle electrical architecture & safety systems
System ArchitectureTractive SystemBrake PlausibilityPowertrainVehicle Integration
As Chief Engineer of CSU's first electric car, I owned the full vehicle architecture and supported the powertrain and suspension teams, managing timelines across six mechanical and electrical subteams to deliver the first FSAE EV prototype in the Rocky Mountain region.
Formula SAE Electric Vehicle System Architecture
Full HV, LV & safety architecture, bench-testable before integration
Context
- Integrated HV powertrain and LV controls into one vehicle-level design meeting FSAE shutdown and safety requirements.
- Coordinated interfaces across inverter PM100DX, Orion VMS2, accumulator, and vehicle controls with mixed signal types.
- Needed a bench-testable architecture to validate interlocks and communication before accumulator integration.
Technical Approach
- Engineered full HV, LV, and safety system architecture and interfaces.
- Defined shutdown, interlock, and communication paths (CAN, analog, digital).
- Created complete KiCad schematics from component datasheets.
Results
- Delivered complete electrical schematics used by all subteams.
- Enabled bench testing of HV, LV, and safety interactions before vehicle integration.
- Reduced integration risk during first high-voltage power-up.
Tractive System Active Lights
Galvanically isolated HV/LV PCB, 60V threshold detection & 555 blink timing
Context
- FSAE rules require visible HV indication: solid red above 60V DC (live), blinking green below 60V (safe), incorrect indication is immediate disqualification.
Approach
- Split the PCB into galvanically isolated HV and LV sections with a 0.3mm creepage barrier, isolated DC-DC converter, and optocoupler relay; HV side uses a resistor divider for comparator threshold detection, LV side drives a 555 timer for the green blink and red enable logic.
Results
- Delivered KiCad schematic and PCB layout with clear HV/LV boundary marking, comparator-based 60V detection, and 555-based blink timing meeting FSAE visibility and frequency requirements.
Brake System Plausibility Design
Hardware brake/throttle conflict detection, latched shutdown, all fault modes tested
Background
- FSAE rules require hardware-based detection of simultaneous brake and throttle, if accelerator exceeds threshold while braking, the system must open shutdown and latch the fault until manual reset; software-only solutions are prohibited.
Technical Approach
- Partitioned the circuit into independent brake and accelerator monitoring sections, each with LM393 comparators checking voltage bounds and signal-loss via pull-up resistors.
- Combined fault conditions through logic gates (74AHC1G02) feeding a latched relay driver (TLP241A); included test points (TP1 to TP9) for calibration and fault injection.
Results
- Completed schematic, PCB layout, and 3D model in KiCad; bench-tested all fault modes confirming correct latch behavior and that shutdown opens within timing requirements.
Battery Pack Structure
First CSU EV accumulator, 180V, 4.1kWh pack, 50 pre-made modules
Background
- First EV accumulator for CSU required integrating 50 pre-made 8p modules (CIE Solutions Lithium Block, Molicel P28A) into a structural container meeting FSAE flame-retardant, isolation, and serviceability requirements, no prior team knowledge existed.
Technical Approach
- Collaborated with a mechanical engineer to define the container: 1018 cold-rolled steel welded enclosure, Garolite G10 interior for HV isolation, laser-cut acrylic module cases, and 9200FR flame-retardant epoxy for bonding.
Results
- Delivered a functional 180V, 4.1kWh pack integrated with Orion BMS 2 and safety electronics; directly informed next-gen 90s5p pack design decisions.
Electrical Powertrain Set-Up
PM100 bench validation, CAN fault-injection framework, thermal derating
Background
- Before integration, the motor controller (PM100) and motor must be validated on bench to verify startup sequencing, fault responses, thermal behavior, and CAN communication, an untested HV powertrain in-vehicle risks damage to drivetrain and safety systems.
Technical Approach
- Built a benchtop test stand with isolated mounting, an E-stop killing both LV and HV, a precharge circuit with manual contactor sequencing, and an independent cooling loop; developed a systematic startup procedure (LV → CAN active → precharge → main contactor → zero-throttle verify → incremental torque).
- Created a CAN test framework to inject faults (CAN dropout, throttle loss, simulated overtemp, undervoltage) and verify controller behavior; logged DC bus voltage, temps, torque command vs. actual, and coolant delta-T.
Results
- Validated a repeatable startup sequence, confirmed all fault-injection cases trigger expected shutdown, and established thermal derating thresholds; system ready for vehicle integration with a known operating envelope.
04Intrepid IC
Combustion car, full harness integration & wireless telemetry
Wiring HarnessCAN BusHaltech ECUTelemetry30+ Sensors
On the combustion car, our two-person electrical team delivered the full vehicle harness and a wireless telemetry unit, the foundation that later grew into sensor fusion and autonomous work.
Vehicle Wiring Integration
Haltech Nexus R5 + 30 sensors, full harness mapped in RapidHarness
Background
- FSAE combustion vehicle requires integration of an ECU (Haltech Nexus R5), transmission control unit, custom Raspberry Pi display, and 30+ sensors into one harness, wiring errors cause no-start, sensor faults, or fire risk.
Technical Approach
- Mapped complete electrical architecture in RapidHarness: wire routing by zone, harness segment lengths, connector pinouts (Deutsch, DTM), and color-coded circuits; documented every ECU input.
- Defined a CAN bus architecture connecting ECU, TCU (servo clutch, solenoid shift), and Raspberry Pi display; mapped CAN message IDs.
- Integrated vehicle dynamics sensors (linear pots, wheel speed, throttle, brake pressure) with headroom for future strain gauges and steering torque sensors.
Results
- Delivered complete wiring documentation enabling repeatable harness builds; established the CAN protocol map used by the telemetry system; harness supports trackside serviceability with labeled service connectors.
Telemetry Control Unit
Teensy + XBee wireless telemetry, ECU + IMU broadcast at ~300m range
Background
- Real-time vehicle data was needed at pit for driver coaching and fault diagnosis; a wired connection was impractical, requiring a wireless system with ~300m range and a meaningful update rate.
Technical Approach
- Selected Teensy for CAN support and speed to parse Haltech ECU broadcasts at 500 kbps; designed power regulation stepping 12V to 3.3V/5V rails with filtering for high-EMI noise immunity.
- Integrated a BMI270 6-axis IMU for yaw rate and acceleration; implemented an XBee 2.4GHz radio (~250 kbps, ~300m range); architected for future sensor fusion and autonomous development.
Results
- Delivered a functional telemetry PCB (EasyEDA schematic, fabricated board) broadcasting ECU and IMU data wirelessly; documented range/bandwidth tradeoffs vs. LoRa and cellular.
05Driverless
Perception, anomaly detection & steering control in the Formula Student Driverless Simulator
LiDAR + CameraYOLOXGBoostSAC / PPO
I built and evaluated a simulated driverless racing stack, from cone sensing and color validation to steering policy research. The three walkthroughs below cover each part of that work.
01LiDAR-Camera Sensor Fusion
Fused LiDAR and camera detections for reactive cone following, reaching 6 cm average match accuracy and resolving straight-line weaving from the LiDAR-only controller.
Read the sensor fusion walkthrough02YOLO Color Anomaly Detection
Labeled 129,025 cone detections and trained an XGBoost safety gate that reduced incorrect color labels reaching control by about 91%.
Read the anomaly detection walkthrough03Reinforcement Learning for Steering
Analyzed six Soft Actor-Critic training runs, diagnosed structural failure modes, and developed a PPO training environment for comparison.
Read the steering control walkthrough
06Other Projects
Selected work, embedded systems, automotive cybersecurity & early CSU race cars
A few standout projects beyond my lead Formula SAE roles, from a competition-winning embedded clock and heavy-vehicle cybersecurity to the early CSU race cars where I cut my teeth on harnesses and vehicle instrumentation.
Sophomore Competition Winner
Mechanical seven-segment clock, ESP32, WiFi UI, servo control
Background
- The competition required a functional electromechanical product from concept to prototype; the team chose a mechanical seven-segment clock combining PCB design, embedded firmware, servo control, and wireless connectivity.
Technical Approach
- Supported PCB implementation: ESP32-WROOM selection (integrated WiFi, GPIO for servos, low cost), USB-C power/programming, and overcurrent protection sizing for servo stall.
- Developed a web-based control interface hosted on the ESP32 for setting time and alarm, with servo PWM control to flip mechanical segments.
Results
- Won the sophomore engineering competition; contributed across the full product cycle from component selection to working prototype.
Cybertruck Challenge
Heavy-vehicle cybersecurity, CAN/J1939 analysis, UDS, ~50 selected nationwide
Background
- Selected as one of ~50 participants nationwide for week-long hands-on cybersecurity training on commercial heavy-duty trucks; mentored by engineers from Bosch, PACCAR, Daimler, and NMFTA.
Technical Approach
- Captured CAN traffic with Wireshark to map arbitration IDs to source ECUs across the J1939 protocol.
- Developed brute-force scripts targeting the UDS Security Access seed/key challenge to probe the authentication mechanism.
- Scripted replay and spoofing attacks with crafted CAN frames; tested which sequences triggered state changes vs. plausibility-check rejections.
Results
- Specific findings under NDA; demonstrated automotive network analysis, diagnostic protocol enumeration, and scripted attack methodology on production heavy vehicles.
Endeavour IC
CSU's first combustion car, wiring harness, electronics placement & MoTeC first start
Background
- Endeavour was the first internal combustion car at Colorado State University, with no inherited electrical design to build from.
Technical Approach
- Built the vehicle wiring harness, laid out electronics placement on the chassis, and tested every circuit before commissioning the engine on a MoTeC ECU.
Results
- Successfully started and ran the car, establishing the electrical foundation every later CSU Formula SAE vehicle was built on.
Dauntless IC
Summer strain-gauge campaign, turning Intrepid track data into suspension & powertrain design loads
Background
- The suspension and powertrain teams needed measured load data, not just assumptions, to design Dauntless, so over the summer I instrumented the Intrepid IC car to capture how it was actually loaded on track.
Technical Approach
- Bonded strain gauges to suspension members and driveline components in Wheatstone-bridge configurations, calibrated each channel to engineering units, and logged the bridge outputs synchronized with vehicle CAN data so strain could be tied to cornering, braking, and acceleration events.
Results
- Converted real track strain into peak and fatigue load cases that fed the Dauntless suspension geometry and informed predicted powertrain loads, replacing guesswork with measured design inputs.
Awards & Achievements
Sep 2025
Engineering Feature
The Rocky Mountain Collegian
Profiled for engineering work with CSU Ram Racing.
Mar 2024
ECE 202 Competition Winner
Colorado State University
Won with a custom PCB and embedded system.
Aug 2023
ECE Department Scholarship
Colorado State University
Merit award for academic performance and ECE contributions.
Feb 2023
Academic Achievement Award
NRHH · CSU Chapter
Recognized for a 4.0 GPA.
Aug 2022
International Scholarship
Colorado State University
Merit scholarship for international students.
Jan 2022
Honor Tuition Award
Universidad de Nariño
Tuition waiver for the highest GPA in the cohort.
Jul 2020
Bachiller CONACED Recognition
CONACED · Pasto
Recognized as a top regional high school graduate.
May 2020
Gloria de Martínez Prize
Gimnasio Los Andes
First student in school history to earn a perfect GPA.
Education
Massachusetts Institute of Technology
Coursework · Deep Learning
Jan 2026 · Winter
Selected for the in-person MIT 6.S191 course taught by Dr. Alexander Amini. Covered neural networks, CNNs, RNNs, and transformers.
Stanford University
Visiting Student · Engineering
May – Aug 2026 · Summer
Selected for a competitive technical and entrepreneurial immersion cohort at Stanford.
Colorado State University
B.S. Computer Engineering
Aug 2022 – Dec 2026
Founded CSU's first FSAE Electric program. Embedded systems, signals, controls, and ECE core curriculum.
Universidad de Nariño
Physics
Sep 2020 – Dec 2022
Two years of undergraduate physics starting at 16. Calculus I–III, Linear Algebra, Physics I–II. Transitioned to the U.S. to continue engineering studies.
Contact Me
Get in touch for collaboration.
