Full-Stack Engineer and Data Scientist with nearly 5 years of experience building production-grade AI systems, backend platforms, and real-time computer vision solutions.
Note: GitHub activity/language widgets skew to recent public repos and under-represent private production work.
- Build and deploy end-to-end AI systems across data, model, backend, and product layers.
- Strong in computer vision, multimodal AI, and low-latency edge inference.
- Production-focused engineer with ownership across architecture, APIs, optimization, and delivery.
- Data Scientist (Applied AI / Computer Vision)
- AI/ML Systems Engineer (MLOps, inference, deployment)
- Backend / Full-Stack Engineer (Python, Go, TypeScript)
- Built an end-to-end AI textile sorting stack at Kapdaa (backend + APIs + UI + CV + hyperspectral models).
- Delivered
>90%garment-type classification across15+classes in real-world sorting conditions. - Achieved hyperspectral fibre classification
F1 > 0.90, including trace blend detection at5-10%. - Optimized Jetson/distributed inference pipelines to
30-40 garments/minand up to~2,100 garments/hour. - Improved targeted enterprise backend execution efficiency by
~30%via Go rewrites and distributed-system improvements.
| Project | What I Delivered | Outcome |
|---|---|---|
| AI Textile Sorting Platform (Kapdaa) | End-to-end stack across APIs, backend, CV + hyperspectral models, and edge deployment | >90% garment classification, F1 > 0.90 on fibre detection, 30-40 garments/min throughput |
| Enterprise Network Platform (Jeavio) | Backend features, distributed-system fixes, focused Go rewrites, deployment/security upgrades | ~30% targeted performance gain, major upgrade delivery, ~80% vulnerability backlog resolved |
| OSS Reliability Contributions | Bug fixes and validation work across CLI tooling, ONNX, and research platforms | Multiple active PRs + merged contributions with tests and standards compliance |
- googleworkspace/cli#700 - added API behavior guardrails for Google Tasks due-time handling.
- googleworkspace/cli#699 - fixed OAuth proxy URL reliability and parameter integrity.
- googleworkspace/cli#698 - added missing auth scopes for Contacts and Gmail settings workflows.
- onnx/onnx#7821 - ConvTranspose validation path hardening and regression coverage.
- ResearchHub/researchhub-backend#3280 - targeted bounty notification behavior fix.
Current focus: AI/ML validation, tooling reliability, and production-safe bug fixes with tests.
- Software Engineering:
Python,TypeScript/JavaScript,Go,C++,SQL,Bash - Backend and Distributed Systems:
FastAPI,Flask,Django,REST,GraphQL,gRPC, microservices - AI/ML and CV:
PyTorch,TensorFlow,Hugging Face,OpenCV,ViTs,Hyperspectral Imaging,ONNX - Edge and Optimization:
NVIDIA Jetson,TensorRT,DeepStream,OpenVINO, low-latency inference pipelines - Cloud and DevOps:
AWS,Azure,GCP,Docker,Kubernetes,CI/CD,Jenkins,Ansible - Data and Infrastructure:
PostgreSQL,MySQL,MongoDB,Redis,Kafka,Spark, ETL/ELT
- Knee Ligament Tear Assessment - 3D CNN pipeline for ACL tear detection and grading from MRI.
- Emotion-based Music Player - end-to-end full-stack CV application with real-time emotion inference.
- SFO Passenger Survey Analysis - statistical modeling and visualization of traveler experience factors.
- Email:
jeevanpawar5890@gmail.com - LinkedIn: linkedin.com/in/jeevan-pawar
- GitHub: github.com/jeevan6996
Open to opportunities in Data Science, AI/ML Systems, Computer Vision, and Backend Engineering.