About me
I am an AI Engineer building production-grade intelligent systems at the intersection of machine learning and software engineering. Currently interning at eInfochips – An Arrow Company, where I execute parameter-efficient LLM fine-tuning using QLoRA and LoRA with FP16/BF16 mixed-precision training, and implement transformer architecture internals — attention mechanisms, tokenization pipelines, and model evaluation metrics — in a real semiconductor AI environment.
Previously at Navneet Education Limited, I engineered a Speech-to-Text voice logging system on GCP that eliminated manual warehouse data entry, built an automated invoice OCR pipeline using GCP Vision API for structured billing data extraction, and developed an internal RAG chatbot via Vertex AI Search and Discovery Engine API — enabling natural-language querying over proprietary product catalogs.
My core technical stack spans Python, PyTorch, scikit-learn, Django, FastAPI, and Google Cloud Platform (Speech-to-Text, Vision API, Vertex AI). I hold a certification as an Oracle Cloud Infrastructure AI Foundations Associate and maintain a CGPA of 9.4 in B.E. Information Technology at Lok Jagruti University.
I architect AI systems end-to-end — from data ingestion and model fine-tuning through REST API integration and full-stack deployment — with deep focus on MLOps, NLP, anomaly detection, and Retrieval-Augmented Generation (RAG).
AI Engineer Intern
- Executing parameter-efficient LLM fine-tuning using QLoRA and LoRA with FP16/BF16 mixed-precision training to adapt large language models on domain-specific semiconductor datasets.
- Implementing transformer architecture internals — attention mechanisms, tokenization pipelines, and model evaluation metrics — within a production-grade semiconductor AI environment.
Software Developer Intern
- Engineered a Speech-to-Text voice logging system on GCP, converting spoken product codes and quantities into structured inventory records — eliminating manual data entry for warehouse teams.
- Built an automated invoice OCR pipeline using GCP Vision API, extracting and structuring billing data from physical documents for finance operations.
- Developed an internal RAG chatbot using Vertex AI Search and Discovery Engine API, enabling natural language querying over proprietary product catalogs and internal documentation.


Hackovate'25
AI Hackathon organized by LJ University
Secured 2nd position in a highly competitive AI hackathon featuring top talent from across the region.
We developed Sensor Guard, an innovative AI-driven security solution designed to enhance surveillance capabilities. Our project utilized advanced sensor integration and real-time data processing (trained on dummy data) to detect anomalies with high precision, showcasing our ability to solve complex real-world problems.
Team
Sneh Patel
Teammate
Contact Form
Please contact me directly at shahsarjan968(at)gmail.com or drop your info here.