Gen AI Developer · Coimbatore, India
I design production-minded RAG, multimodal, and agentic applications for real-world workflows, from healthcare document intelligence to AI-powered research tools.
01 / About
I’m an AI and Data Science engineer who enjoys taking fuzzy problems and turning them into reliable interfaces, retrieval pipelines, and decision-support systems.
My work sits at the intersection of language models, structured data, and human context. I care about the parts that make an AI product dependable: evaluation, grounding, cost, latency, and a clear user experience.
Start a conversation02 / Experience
Hands-on experience across healthcare AI, NLP, computer vision, and LLM application development.
Gen AI Developer · Connected Value Health
Built a Batch API pipeline with Mistral OCR for PDF processing under the Patient Nexus EBV system, cutting per-page cost by 25%. Trained a Gradient Boosting model to predict drug coverage status from insurance benefit documents.
AI Intern · Ei Systems
Built MediSaga, a medical RAG chatbot using RAPTOR indexing, Milvus, and quantized Llama2 for efficient CPU-based question answering.
ML Intern · Yanne Technologies
Developed a LangChain + GPT-3.5 medical assistant and an OpenCV checkbox detector for Vizhi Eye Care NGO.
B.Tech in Artificial Intelligence and Data Science · KPR Institute of Engineering and Technology
2022 — 2026 · Coimbatore, India CGPA 8.0 / 10
03 / Selected work
Systems built to answer questions, verify claims, and turn messy inputs into useful actions.
01Multimodal misinformation detection for text and image claims, combining BERT + BiLSTM classification, OCR, live web search, and LLM-generated rationales.
View on GitHub
02Dockerized RAG system that accepts PDF, audio, and video inputs for context-aware answers.
View on GitHub
03Natural language to SQL with schema detection, validation, and real-time result display.
View on GitHub
04An MCP server for AI-driven expense and employee management with Gemini-based analytics.
View on GitHub
05Medical QA chatbot built with RAPTOR indexing, Milvus, and Llama2 for grounded retrieval.
View on GitHub04 / Toolkit
A practical stack for building, evaluating, and shipping intelligent products.
RAG · LLM integration · Transformers · Fine-tuning · BERT · BiLSTM · Gradient Boosting · Classification
Python · FastAPI · FastMCP · Streamlit · Chainlit · React · LangChain · Docker
Milvus · ChromaDB · FAISS · Pinecone · MongoDB Atlas · PostgreSQL · SQLite
OpenCV · YOLOv5/v8 · OCR · Pandas · NumPy · Git · Docker Compose · Bash
05 / Resume
See the full record of my experience, education, projects, technical skills, and awards.
06 / Certifications & awards
Selected credentials and recognition spanning agents, MCP, MLOps, transformers, and applied machine learning.
Udemy
Hugging Face
View credentialAnthropic · Credential ID: 6s4idecta35u
Verify credentialGoogle · Credential ID: 14227411
Verify credentialGoogle · Credential ID: 14227525
Verify credentialDRDO & KPR Institute of Engineering and Technology
KPR Institute of Engineering and Technology · Credential ID: 276420250028/ND
Computer vision, NLP pipelines, Generative AI prototyping, data analytics, and real-time AI applications.
Verify credentialGoogle · Credential ID: 15097460
Prompt engineering, Gemini in Vertex AI, multimodal prompting, image analysis, and response quality tuning.
Verify credentialSector Skill Council Nasscom · Credential ID: FSP/2022/12/4954758
Data collection, storage basics, pandas DataFrame handling, and validation techniques.
View certificateCoursera · Credential ID: VPRE7WUSMFHY
Statistical visualization, dimensionality reduction, cluster-based pattern discovery, and analytical storytelling.
Verify credentialAnalytical thinking, data lifecycle understanding, visualization tools, and data ethics.
Verify credentialCoursera
NumPy operations, pandas manipulation, data wrangling, filtering, and tabular analysis.
Verify credential365 Data Science
Git fundamentals, branching workflows, commit history, diffs, and GitHub collaboration.
Verify credentialIBM
Supervised and unsupervised learning, model evaluation, feature engineering, and scikit-learn.
Verify credential07 / Contact
Let’s talk about AI products that need to be useful, grounded, and ready for the real world.
suryakf04@gmail.com