VSSV Aditya Ram

Aditya Ram · AI engineer

Agents, voice,
useful products.

I’m Aditya, an AI engineer at Tata Consultancy Services. I build enterprise AI prototypes on Azure, working across agents, retrieval and real-time voice.

Outside work, I build and run Budget and this site’s assistant on Cloudflare, taking ideas through to everyday use. I’m open to AI engineering roles.

View selected work Download resume Explore the atlas

Selected work

Enterprise prototypes. Independent products.

Three ways I turn AI into something useful.

01
Client POC · built at TCS · 2026

Telecom customer assistant

Account questions across chat and voice. I built the routing, approved-tool layer and Azure deployment, with LangGraph for questions that need multiple lookups.

Azure OpenAILangGraphRealtime voice / WebRTCAzure AI Foundry & MAF (explored)
QuestionApproved toolsGrounded answer
Explore the case study →
Contribution
Architecture, backend, UI and Azure deployment
Evidence
149 mocked unit tests; 47-check live suite
Scope
TCS client POC; voice in 4 languages
02
Live · budget.vaditya.in

Budget, a daily finance app

A monthly budget becomes a daily allowance. I designed, built and operate the app, including a text-and-voice assistant that asks before changing data.

Cloudflare WorkersD1 (SQLite)Workers AIGemini
Budget’s daily allowance and expense entry screen, shown with sample data
Preview Budget Project page
Contribution
Designed, built and operated end to end
Evidence
Live product I use daily
Boundary
AI-proposed changes need confirmation
03
Live · bottom-right of this page

Ask about Aditya

A portfolio should answer follow-up questions. I built this assistant to answer from my published work, cite sources and say when the evidence is missing.

Cloudflare VectorizeWorkers AI embeddingsGeminiRAG
How it works
Contribution
Retrieval, API and accessible chat UI
Evidence
Try a real question on this page
Boundary
Source-linked answers; model fallback

About

Models interpret. The system stays accountable.

VSSV Aditya Ram

VSSV Aditya Ram

At TCS, I build applied AI systems: clear tool boundaries, grounded answers, and a backend that handles real failure modes.

Outside work, I design and operate Budget and this site’s assistant. Building both end to end keeps me close to the people using them.

Education

B.Tech, Computer Science Engineering

GITAM University, Visakhapatnam

2021 – 2025 CGPA 8.11

Skills

The tools behind the work.

Agentic AI & LLMs

Deciding what the model decides.

Azure OpenAI (GPT-5 family)Azure AI FoundryMicrosoft Agent FrameworkHosted agents

Azure platform

The services under the last year's work.

Azure VMs (Ubuntu)Cosmos DB for NoSQLAzure AI SearchRealtime API

Retrieval & data

Grounding answers in the right thing.

RAGEmbeddings & vector searchCloudflare VectorizeSemantic search

Backend & edge

Small services that stay up.

PythonFastAPITypeScriptCloudflare Workers

Vision & classic ML

Where I started.

OpenCVMediaPipeYOLO (v8 / v11)EfficientNet

Product & web

Shipping the thing people touch.

Vanilla JSPWAWeb Speech APIStreamlit
Explore the detailed stack
Tools in practice
VSSV Aditya Ram | AI Engineer Four layers stacked top to bottom: models, agents and orchestration, data and retrieval, and runtime and edge — the same four groups listed below, connected by one rail. Models GPT-5 family · Realtime voice · Gemini Agents & orchestration AI Foundry · Agent Framework · LangGraph Data & retrieval Cosmos DB · AI Search · Vectorize + D1 Runtime & edge Azure VM · Cloudflare Workers · FastAPI
Four layers, top to bottom — tools used across my projects; each case study identifies its own stack.

Models

  • Azure OpenAI · GPT-5 familyrouting, planning, grounded answers
  • Realtime voice + live transcriptionWebRTC, tool calls mid-conversation
  • Gemini + Cloudflare Workers AIstructured output, automatic fallback

Agents & orchestration

  • Azure AI Foundryprompt agents, OpenAPI tools
  • Microsoft Agent Frameworkhosted and standalone agents
  • LangGraphplan → fetch → answer graphs
  • Deterministic-first routingconfidence gating, clarification

Data & retrieval

  • Azure Cosmos DB (NoSQL)conversation history
  • Azure AI Searchfull-text over conversations
  • Cloudflare Vectorize + D1embeddings, SQLite at the edge
  • RAG with grounded citations

Runtime & edge

  • Azure VMs · nginx · systemdTLS via Let's Encrypt
  • WebRTC + self-hosted TURNvoice through locked-down networks
  • Cloudflare WorkersAPI + PWA from one origin
  • FastAPI · Python · TypeScript
Tools used across client prototypes and independent live products.

Experience

From data to applied AI.

Feb 2026 – presentHyderabad

Systems Engineer – Data Science (AI/ML)

Tata Consultancy Services

  • Built client POCs and prototypes for telecom customer service, incident response, and sales/service assistance. Explore the TCS work.
  • Owned the telecom assistant from routing and real-time voice to Azure deployment: 149 mocked unit tests and a 47-check live regression suite.
  • Also worked on PPE detection trained on 45,000+ images and healthcare-inventory workflows with FEFO logic and role-based access.
May – Aug 2024Hyderabad

Trainee Intern, Machine Learning

Tech Mahindra

  • Built an IoT digital-twin framework for real-time sensor data on an edge–cloud architecture, with a BLE Android app for sensor communication.
  • ML gesture recognition (Namaskar detection) with OpenCV and MediaPipe; improved responsiveness and real-time detection accuracy.
Aug – Oct 2023Remote

Data Analyst Lead Intern

Ozibook

  • Led a team of 35 interns; social media engagement up 20%. Power BI dashboards for client reporting; Selenium automation that removed most manual data extraction.
Certifications & recognition Claude foundations, Gemini Enterprise & Google Cloud skill badges

Anthropic

Claude Certified Architect — Foundations

Certified on designing agentic systems around Claude: tool use, context management, and production deployment patterns.

Anthropic

Claude Certified Developer — Foundations

Certified on building with the Claude API and Claude Code: prompting, tool integration, and agentic workflows.

Runner-up

TCS Internal Hackathon

Retail sentiment and product-intelligence system — Streamlit, Altair, OpenAI API, LangChain, RAG.

Google Cloud

Certified Partner Specialist — Gemini Enterprise Deployment

Partner certification in deploying Gemini Enterprise for organizations.

Google Cloud

Certified Partner Specialist — Gemini Enterprise Agent Development

Partner certification in building agents on Gemini Enterprise.

Champion

LIREL Champion, TCS

Recognised for performance and consistent execution among peers.

Top 5 on campus

PwC Launchpad

1.5-year programme in Python fundamentals and the Salesforce platform.

Contact

Open to AI engineering and agentic-systems roles.

Email or LinkedIn is fastest. The form opens your mail app with everything filled in, so you can read it before it goes.

After hours

A little room for play.

Co-op puzzles, quick races and Mafia. Bring a friend.

Enter the arcade

Hidden

A door with no sign.

If it's yours, you already have the key.