Sep 2026 — Now
AI Engineer
Straive (formerly SPi Global)
- Building generative AI and NLP solutions for content, data and learning workflows at a global content-technology company.
Hi, I'm
Durgaram
AI Engineer at Straive. I build LLM, RAG, speech and vision systems — and ship them into the hands of the people who use them every day.
01 — About
I turn messy, real-world problems into *AI *that *ships — not demos, but tools people rely on every day.
I'm an AI/ML engineer who builds the whole thing — fine-tuned LLMs, RAG pipelines, agentic workflows, speech recognition, OCR and computer vision — with LangChain, PyTorch, Hugging Face and FastAPI, then ships it with Docker and CI/CD.
Before Straive, I spent a year inside a hospital building clinical AI: discharge summaries written by a fine-tuned LLM, a medical RAG assistant that cites its sources page by page, and transcription of doctor–patient conversations. Working next to doctors taught me to start from the problem, not the model.
NowAI Engineer at Straive (formerly SPi Global)
BeforeClinical AI at MAPIMS hospital
StudiedB.Tech, AI & Data Science Anna University · 2022–26 · 82%
FocusLLMs · RAG · Agents · Speech · Vision
BasedCuddalore, Tamil Nadu IST · UTC+5:30
15+Projects shipped
5Roles in AI
1Publication
01 — Listen
Start from the problem
Requirements, SRS and user stories first. I sit with the people who'll use it until the real need is clear.
02 — Build
Own it end-to-end
Data, model, API and interface — one person who understands every layer, so nothing gets lost between them.
03 — Ship
Make it hold up
Docker, CI/CD and citations you can check. Done means people use it every day, not that the demo worked once.
02 — Experience
Where I've worked
Jan 2026 — Jul 2026
AI & Software Engineer
MAPIMS — Melmaruvathur Adhiparasakthi Institute of Medical Sciences
- Shipped a discharge-summary product: a fine-tuned LLM that turns patient notes into structured clinical reports.
- Built a medical RAG chatbot over hospital documents with LangChain, ChromaDB and page-level citations.
- Developed a clinical ASR workflow for patient–doctor conversations, and an NLP-routed ticket management system.
- Deployed AI services via FastAPI/Flask and Docker into staff-facing hospital tools.
Jun 2025 — Dec 2025
AI Trainee
Transvaal Global Tech
- Contributed to an AI-based HRMS covering records, attendance, leave and payroll.
- Built NLP/LLM features for HR document automation, policy Q&A and employee self-service.
Feb 2025 — Mar 2025
Machine Learning Intern
Prodigy InfoTech
- Trained and evaluated classification models with Scikit-learn and Pandas; built preprocessing and feature-engineering pipelines.
Aug 2024 — Sep 2024
Machine Learning Intern
TRIOS Technologies
- Contributed to predictive-analytics models: preprocessing, feature selection and evaluation.
03 — Selected work
Things I've built
Fully offline conversational avatar with Wav2Lip lip-sync. Qwen LLM, Whisper ASR and TTS behind FastAPI, rendered in Three.js.
A learning management system that guides learners through structured career paths with progress tracking.
IndexStack
01
Cascade Failure Prevention Agent
Multi-agent system that flags power-grid failures before they cascade, with graph RAG over fault history.
02
Neural OCR Engine
Layout-aware OCR on scanned PDFs, parsed into custom XML schemas with a review UI.
03
Clinical ASR
Patient–doctor conversation transcription with review rules for clinical documentation.
04
05AI Feedback System
Voice capture to structured LLM summaries, themes and action items.
Auto Image Trainer Studio
Annotation, dataset management and model training in one web platform.
06Vehicle Monitoring System
Real-time vehicle detection and tracking for traffic surveillance.
07Face Recognition Attendance
Anti-spoofing attendance on Raspberry Pi — the basis of my publication.
08YOLO Trainer & Deployer
Train, fine-tune and deploy YOLO detectors from a Streamlit app.
09Hand Gesture Mouse
Control the cursor with hand gestures through a webcam.
10Multi-Disease Prediction
Early detection of multiple diseases from symptoms and patient data.
11Brain Tumor Prediction
Deep learning detection of brain tumors from MRI scans.
12Movie Sentiment Analysis
NLP app that reads reviews and gauges audience reception.
13Smart AI Calculator
Solves math written in plain language.
14Handwritten Digit Recognizer
CNN trained on MNIST with TensorFlow.
More experiments on GitHub ↗
04 — Skills
My neural stack
Drag nodes · click to explore
05 — Credentials
Research & certifications
Publication · IIP Series
Face Recognition Based Attendance System Using Anti-Spoofing Methods
E-ISBN 978-93-6252-750-9
Print ISBN 978-93-6252-009-8
Master Claude Code & Model Context ProtocolAnthropic Academy
Google Cloud Data Analytics · Cloud EngineerGoogle Cloud
Complete Data Science, ML, DL & NLP BootcampUdemy ↗
Advanced Learning AlgorithmsCoursera ↗
Supervised ML: Regression & ClassificationCoursera ↗
GenAI Job SimulationBCG X ↗
Data Analytics & VisualizationAccenture ↗
Python for Machine LearningGreat Learning ↗