Pranav Chetan

Hi, I'm

Pranav Chetan

Software engineer · Applied ML + backend · Open to roles

I build AI for real hardware, which mostly means convincing expensive machines to do what my code says. So far I've talked a robot into driving itself, a laser scanner into speaking Linux, a chatbot into reading Congress, and an F1 model into guessing Sunday's winner. The robot and I are on good terms now. Mostly.

See my projects About me
A

Interface

React, TypeScript, Node.js, REST APIs

B

Languages

Python, C / C++, Java, Go, SQL

C

Models

PyTorch, reinforcement learning, gradient boosting

D

Retrieval

RAG pipelines, tool-calling LLM agents

E

Data

SQL, Pandas, Medallion architecture

F

Systems

Linux, ROS 2, Jetson, USB protocols

Exploded view

What's inside

Six layers, from the interface down to the operating system. Every project below uses most of them.

Drawn by
Pranav Chetan
Education
B.S. CS · Wayne State '25
Status
Open to roles

About me

Who's building this

I like problems where the model is only half the job. The other half is the data it stands on, the metric that proves it works, and the service that puts it in front of people.

I studied computer science at Wayne State University on the GM-sponsored project track. I've built a self-driving robot stack, a 3D-scanning workstation, ranking models and retrieval pipelines, and I care most about being able to prove a system works.

See the projects
Education
B.S. Computer ScienceWayne State University · Class of 2025
Focus
Applied ML, robotics, backend services, Linux systems
Certifications
Machine Learning SpecializationStanford & DeepLearning.AI · 2024
Looking for
Software, backend and systems engineering roles

Selected work

Projects I've built

Four projects · 2023–2026

  1. 01 · Robotics AMR An autonomous mobile robot that maps a space, plans its own routes and navigates cluttered rooms to run inventory missions. Reaches every goal in 96.5% of test runs, and 94.5% of runs finish without touching a thing.
  2. 02 · 3D scanning CloudClean A workstation that turns laser scans into measured, inspection-ready parts. Import a scan and Autopilot does the rest. Reverse-engineered the scanner so it runs on Linux, then cut its scan error about 10×.
  3. 03 · GM sponsored Policy analytics An AI analyst that answers questions about bills, votes and sponsors in plain English, running entirely on device. Turned minutes or hours of SQL and BI work into answers in seconds, each traced to its source.
  4. 04 · Machine learning F1 prediction A model that predicts Formula 1 finishing orders from 10+ seasons of race data. Scored 10–25% better on ranking metrics than accuracy-only baselines.
Robotics Autonomy + AI · 2026 · In active development

AMR: autonomous inventory robot

Autonomous mobile robots are taking over the walking in warehouses, factories and hospitals. AMR is my full-stack build of one: a tracked robot that maps a space with LiDAR, knows where it is, plans its own routes and drives through cluttered, changing rooms, with a safety layer that always has the final say.

I'm building the software at every layer, from mapping and route planning to the AI that drives it and the web console that dispatches it. Try the console: send it a goal, drive it yourself, or drop an obstacle in its path.

Use casesInventory counts · drive-to-spot scans · box inspection · shelf audits · 3D room maps · aisle patrols

Goal success96.5%

Test runs where the robot reached every goal it was given

Obstacle avoidance94.5%

Test runs finished without touching a single obstacle

200 test runs in a cluttered room with partitions and alcoves

  • Python
  • PyTorch
  • Reinforcement learning
  • ROS 2
  • FAST-LIO2 SLAM
  • Jetson Orin
  • LiDAR
  • RealSense
  • YOLO11
Read the full story Watch the real robot · 0:51
3D scanning Reverse engineering + full stack · 2026

CloudClean 3D scanning workstation

A workstation that turns raw scans from a precision laser scanner into cleaned, measured, inspection-ready parts, all from a web browser, with an AI assistant that can run the whole app.

AutopilotYou: import a scan

  1. Import
  2. Clean
  3. Merge
  4. Mesh
  5. Inspect
  6. Export

You: 1 stepAutopilot: the other 5, no clicks

Import a point cloud and that's the whole job. Autopilot cleans it, checks and merges multiple scans, builds the mesh, inspects the part against its CAD model if you've added one, and exports it, without you clicking a single button. In most scanning software, each of those is a separate tool you run and tune by hand.

The scanner's own software doesn't run on Linux, so I worked out how it talks over USB and wrote my own driver.

~0×Lower scan error than my first driver
0 fpsLive capture on Linux
~0App tools the AI assistant can use
~0Automated tests
  • Python
  • Open3D
  • OpenCV
  • FastAPI
  • React
  • TypeScript
  • three.js
  • vLLM
Read the full story Code on GitHub
GM sponsored Applied AI + full stack · 2025

Public policy analytics

A conversational AI system over congressional and legislative data. Policy analysts ask questions in plain English and get answers in seconds, with no SQL or BI tools involved.

It all runs on device: the language model, the search and the data live on one machine, so no question or answer is ever sent to the cloud.

Try an example question

 

Bill recordRoll-call voteGold · votes
0%Answers grounded in Gold tables
Min → secQuestion turnaround
  • Python
  • RAG
  • LLM
  • SQL
  • Medallion architecture
  • Audit logging
Read the full story
Machine learning 2023

F1 Grand Prix winner prediction

A race-outcome and driver-ranking system trained on more than ten seasons of Formula 1 results. It's scored the way a ranking should be, with NDCG and MAP, not raw accuracy.

Live

Next raceLoading the calendar…

Predicted winner…

Bet against the model
0Seasons of race data
0–0Features per race entry
0Model families compared
+0–0%NDCG / MAP lift over baseline
  • Python
  • scikit-learn
  • Pandas
  • Random forest
  • Gradient boosting
  • DNN
Read the full story
Lap pace
1.0×

Contact

Hiring for a software role?

I'm looking for full-time software, backend and systems roles. Email is the fastest way to reach me.

pranch555@gmail.com
Connect on LinkedIn View resume