Post Job Free
Sign in

AI/ML Engineer - LLM & Computer Vision

Location:
Indian Landing, LA
Posted:
November 16, 2026

Contact this candidate

Resume:

Sagar Gyanchandani

+91-773******* ***************@*****.*** linkedin.com/in/mr-sagar-gyanchandani Professional Summary

AI/ML Engineer delivering $1M+ impact via production ML systems, LLM applications, and intelligent automation. Architected multi-agent AI solutions (Llama, GPT-4) processing 10K+ daily transactions at 99%+ accuracy. Computer Vision, MLOps, and enterprise AI expertise. B.Tech CSE 4+ years Deloitte, Siemens. Technical Skills

Generative AI & LLMs: GPT-4, Llama 3.1, LangChain, LlamaIndex, Hugging Face (Transformers, PEFT, TGI), RAG, Prompt Engineering, Fine-tuning, ChromaDB, Milvus DB (Vector Search) ML/DL Frameworks: TensorFlow, PyTorch, Scikit-learn, XGBoost, Keras, ONNX (Model Optimization), Transfer Learning, Feature Engineering

Computer Vision: OpenCV, YOLO v5, RetinaFace, ArcFace, OCR, Whisper, Nvidia Deepstream, Real-time Inference Cloud & DevOps: AWS (S3, EC2, Glue, Lambda), GCP, Azure (OpenAI API), Docker, Kubernetes, CI/CD, MLOps Pipelines

Backend & Data: Python (Advanced), FastAPI, SQL, RPA, Dataiku DSS, Data Engineering Core: CUDA, Linux, Agile

Professional Experience

Deloitte Jun 2025 – Present

AI and Data Science Engineer II Gurugram Jun 2026 – Present Consultant Gurugram Jun 2025 – Jun 2026

FedEx: Intelligent Workload Allocation & Document Automation Azure OpenAI, RAG, Python, RPA, MLOps

– Architected intelligent intake-to-routing system for CA1 payment disputes: deployed skill-matching ML model dynamically allocating 5,000+ monthly items; 65% faster resolution (5 days 48 hrs), $200K+ annual cash flow acceleration.

– CopyDocs RAG automation: Engineered document retrieval system using Azure OpenAI embeddings + intelligent parsing; processed 200+ daily emails with 92% cycle-time reduction (2 days 15 mins), 98% accuracy, eliminated 300+ hrs/quarter manual work. Marine Atlantic Inc: Production Multi-Agent LLM System Llama 3.1-70B, LlamaIndex, RAG, Milvus DB

– Deployed production LLM system processing 2,000+ monthly marine operations queries with 96% accuracy, sub-2-sec latency using LLaMA 3.1-70B; engineered Milvus vector DB for semantic context retrieval.

– Multi-agent orchestration (parallel extraction, routing, synthesis); 80% faster resolution (4 hrs 3 mins); 99.2% uptime across 500K+ monthly transactions.

Siemens Jan 2025 – Jun 2025

AI ML Developer Pune

DMI Screen Classification Engine Custom CNN, Transfer Learning, ONNX Optimization

– Trained custom CNN from 50K+ labeled images achieving 99.2% precision; optimized to ONNX format for edge deployment on Jetson devices.

– 100+ images/sec throughput; eliminated 20+ hrs/week manual inspection, enabled 24/7 automated workflows, 15% production downtime reduction.

OCR Fine-Tuning & Verification Pipeline PyTesseract, Fine-tuning, Model Optimization

– Fine-tuned OCR model: 82% 96.5% accuracy via domain-specific training; engineered ML-based confidence scoring + exception flagging reducing manual validation by 60% while maintaining 99%+ production correctness.

I2E Consulting Oct 2023 – Jan 2025

Senior Software Developer (Data Scientist) Remote (Mumbai) Pfizer GenAI Enterprise Chatbot GPT-3.5, LangChain, RAG, AWS Glue, Data Engineering

– Deployed production LLM chatbot serving 500+ internal users, 3,000+ monthly queries at 94% accuracy; engineered RAG pipeline with AWS Glue + S3 ETL for real-time data fusion.

– 70% latency reduction; 40% fewer support tickets, $150K+ annual labor savings. Video-to-Question Generator LLaMA 3, Whisper, RAG, ChromaDB, Prompt Engineering

– End-to-end ML pipeline: audio text (Whisper 98.5% WER) semantic questions (LLaMA 3 + prompt engineering) RAG retrieval (ChromaDB). Processed 1,000+ videos, generated 50K+ exam questions for 2,000+ students.

– 85% productivity gain: 8 hrs/video 12 mins/video; scaled content creation across e-learning platform. Pathology Abnormality Detection PyTorch, Transfer Learning, Feature Engineering

– Deep learning pipeline (transfer learning + custom feature engineering) on 10K+ high-resolution pathology slides; 94% sensitivity, 91% specificity.

– Integrated LIS alerts for automated prioritization; 30% faster pathologist review cycle. AI MESH Generative AI Platform Dataiku DSS, Azure OpenAI, RAG, Data Engineering

– Deployed enterprise LLM platform (Azure GPT-3.5) for 3 business units; engineered RAG pipeline with intelligent PDF processing, OCR, and semantic search across 50K+ documents.

– 11x latency improvement (45 min 4 min/file); 200+ concurrent users; enabled cross-functional GenAI adoption.

Auriga IT Consulting Private Limited Oct 2021 – Oct 2023 Associate Software Engineer Jaipur, Rajasthan

AgriBot: Government AI Advisory System GPT-3.5, FastAPI, Feature Engineering, J&K Government

– Semantic similarity ML engine (300+ agricultural facts) + GPT-3.5; 12,000+ monthly active farmers, 99.5% uptime, sub-500ms latency.

– 92% accuracy validated on 1,000+ real-world interactions; government pilot achieved 15% yield improvement across region.

ATMS: Real-Time Road Safety System YOLO v4, Nvidia Deepstream, Computer Vision, NHAI Government

– Real-time video analytics across 50+ highway toll booths: YOLO v4 + Deepstream pipeline detecting wrong-way vehicles, congestion, speed violations at 97% precision.

– 18x faster incident response (8 mins 45 secs); prevented 200 accidents/year; processed 5,000+ detections/hour with <100ms latency.

Face Recognition & Identity Verification RetinaFace, ArcFace, ONNX, Computer Vision

– Production-grade face recognition (RetinaFace detection + ArcFace embeddings + similarity matching) achieving 99.1% accuracy on 100K+ identity database; integrated robust angle/mask/occlusion detection.

– 500K+ daily requests, 150ms P95 latency; powered enterprise physical security system across multiple facilities. Education

B.Tech in Computer Science Engineering GPA: 8.95/10 2018 – 2022 Rajasthan Technical University

Achievements

• Hackathon Champion: Auriga IT Techathon (2x); technical blogger on Docker, face recognition, LLM deployment.



Contact this candidate