Building production ML systems that actually ship — from edge devices to cloud APIs.
GenAI Engineer Intern @ AllCognix AI, working on RAG pipelines, LLM systems, and production ML infrastructure.
- Authoring a research paper on edge-AI deployment for agricultural CV — benchmarking TensorRT, ONNX Runtime, and OpenVINO on Jetson Nano.
- Cut RAG latency by 40% and token costs by 60% via Haystack 2.x migration
- Reduced document ingestion from 70s → 27s with parallel processing
- Deployed wheat disease classifier: 92% accuracy, 75% model compression via ONNX + TensorRT
- Trained LightGBM on 5.5M rows with spatial-temporal feature engineering
ML / Deep Learning
MLOps & Deployment
Backend & Data
|
Research-focused study on deployment-aware agricultural AI for constrained edge hardware. Investigates INT8/FP16 quantization stability across TensorRT, ONNX Runtime, and OpenVINO, introducing a novel Deployment Efficiency Score (DES) metric balancing accuracy and throughput.
54.5 FPS edge inference · INT8 recovery engineering · leakage-audited benchmark |
Production-style ML platform for wheat disease diagnosis featuring ConvNeXt-Tiny inference, CLIP-based validation, OpenCV symptom overlays, GPT-powered recommendations, and human-in-the-loop feedback collection.
88.46% accuracy · 75% model compression · live deployed system |
|
Geospatial ML on 55M rows with Haversine distance features. Containerized and served via FastAPI.
55M rows · containerized API |
LightGBM pipeline on multi-store retail data with automated feature enrichment and PostgreSQL logging. Deployed on Render.
Automated pipeline · production-deployed |
- Building an end-to-end MLOps pipeline with drift detection + automated retraining
- Learning:
Prefect·Evidently AI·Prometheus·Kubernetes - Authoring a research paper on edge-AI deployment for agricultural CV — targeting CEA + arXiv cs.CV
- Open to ML Engineer / GenAI internship roles — India & remote international
Pune, India · B.Tech CSE (AI & Analytics) · MIT ADT University · 2028