AI/ML enthusiast and full-stack developer
Cyril Sabu George
Integrated M.Tech student at VIT Vellore, currently working as an AI Intern at FreshToHome on demand forecasting systems that turn retail data into planning signals, inventory decisions, and more reliable operations.
Demand forecasting
Multimodal AI systems
Full-stack delivery
Operational intelligence
I work across data, backend, and interface layers, translating model behavior and product ambiguity into workflows people can actually trust and use.
The work includes forecasting pipelines, agentic systems, computer vision products, and the less glamorous details that make them hold up under production constraints.
Selected work
Systems that move from notebook to interface.
Current role
AI Intern, FreshToHome
Working on demand forecasting systems that prepare retail data, compare model behavior, and surface practical signals for planning, inventory movement, and substitution decisions across fast-moving operations.
Current
AI internship focused on demand forecasting at FreshToHome.
9.4
CGPA at VIT Vellore, Rank 6 among 200+ students as of semester 6.
500+
Participants impacted through AI/ML track integration across events and workshops.
AI systems
Multimodal RAG and agentic workflows
Built production-facing AI systems with retrieval, voice input, text-to-speech, approval flows, and measurable real-time performance across backend and UI surfaces.
Leadership
Technical programs with external reach
As External Affairs Head at ADG VIT, I coordinate industry-facing outreach and help bring stronger AI/ML content into hackathons, workshops, and community technical programs.
Resume highlights
The formal version, kept readable.
Summary
Master's student in Computer Science at VIT Vellore delivering production-grade AI systems with strong ownership, accountability, and practical full-stack execution under real constraints.
Education
Vellore Institute of Technology, Vellore
CGPA: 9.4/10
Rank 6 out of 200+ students as of Semester 6
Experience
FreshToHome
Building demand forecasting workflows for retail planning, feature preparation, model comparison, and operational decision support.
Experience
GlobalLogic
Engineering multimodal RAG systems with voice input, text-to-speech, hybrid retrieval, reranking, caching, and production-ready modular pipelines.
Leadership
Advanced Developers Group, VIT
Lead industry partnerships and technical outreach for flagship events, bringing stronger AI/ML programming to 250+ participant experiences.
Core skills
- Python, JavaScript, C++, SQL
- FastAPI, React, Flask, LangChain, LangGraph
- TensorFlow, PyTorch, OpenCV, RAG, LLM orchestration
- PostgreSQL, Redis, Firebase, AWS, GCP, Azure, Docker, Kubernetes
Selected projects
- Astra: multi-agent assistant platform with 11 API routes and sub-1.5s key flows
- Scalable RecSys Engine: recommendation system designed for 25M+ records
- Fridge Vision: YOLO-based image-to-recipe pipeline with 67 food classes
Working range
The stack spans modeling, systems, and interface delivery.
Cleaning, features, forecasting, evaluation, metrics, and model sanity checks.
React interfaces, responsive pages, accessible components, and useful product surfaces.
APIs, integrations, testing, refactors, debugging, and maintainable delivery.
RAG pipelines, transfer learning, computer vision, and orchestration for real workflows.
Process
A tight loop, with room for taste.
How I keep momentum
Small loops beat grand rewrites.
I treat every build like a live system: read the signals, choose the smallest useful move, then verify the result before adding more complexity.
-
Observe
Read the current system, visible UI, data shape, and user path.
-
Frame
Find the real constraint and decide what improvement matters most.
-
Build
Move in small source changes that preserve coherence and states.
-
Review
Refresh the surface, compare the before and after, then tighten.
Contact
For internships, collaborations, or product builds, reach out directly.
Based in VIT Vellore and currently working on AI systems for production use cases, especially forecasting, agent workflows, and full-stack AI products.