Hello World! I'm
Ansuman
AI Engineer && Problem Solver
I build systems where AI feels at home and servers don’t complain.
Exploring the space between intelligent models and solid infrastructure,
refining my skills to craft solutions that are both smart and scalable...

// Fun fact: I debug with println() and I'm not ashamed of it

// Another one: My code is like my coffee - strong, reliable, and occasionally bitter

/* about_me.go */

package main


type Engineer struct {

name string

location string

role string

}


func NewEngineer() Engineer {

return Engineer{

name: "Ansuman",

location: "Bengaluru, India",

role: "AI Engineer",

}

}


About Me

Currently working at Vedantu, where I get to build AI systems that recommend questions and content for millions of learners. My projects usually involve multi-agent setups, RAG pipelines, and fine-tuned models, but at their core, they’re just me trying to turn ideas into things that actually work.

My approach to engineering is the same as when I first picked up an Arduino at 11—curious, experimental, and a little reckless. I still break things just to see how they work, and I still end up staying up too late debugging why something isn’t behaving the way I thought it would.

// Fun fact:I write such good code that my rubber duck debugging sessions turned into TED talks.

Core Technologies:

Go
Python
JavaScript
TypeScript
Java
C++

Technical Stacks

// The tools I debug with

Programming Languages
const languages = [
Go
// My go-to for performance
Java
// For when I miss semicolons
Python
// Because readability counts
C++
// For memory lane trips
JavaScript
// Because everything is an object
];
Development Stack
Multithreading
TCP/IP
HTTP
UDP
Operating Systems
Cloud Computing
Databases
Networking
FastAPI
Django
REST APIs
Microservices
HTML
CSS
DOM Manipulation
React
React Native
Flutter
Swift UI

// Full-stack at heart: from TCP/IP to centering divs

Tools & DevOps
Go-Gin
Django
Spring Boot
Express
Docker
PostgreSQL
MongoDB
AWS
GCP
Azure
Kubernetes
Jenkins
Git
Redis
Vector DBs
SQLAlchemy
Pinecone
Weaviate
AWS (SageMaker, Lambda, EC2)
Google Cloud (Vertex AI)
Azure (ML Studio)

// I'm a tool-agnostic developer

Generative AI
RAG
RAGAS
LangChain
LangGraph
AI Agents
Multi-Agent Systems
MCP
Fine-Tuning
LangFuse
LangSmith
Unsloth
PyTorch
TensorFlow

// From RAG to production-grade agents

Work Experience
$ cat career_history.log
ai_engineer.service.ts
AI Engineer
@Vedantu
Jul 2025 – Present
  • Built Multi-Modal RAG pipeline (Weaviate + CLIP) for 3M+ questions → boosted recommendations by 70% for 30M+ users
  • Added RAGAS evaluation for automated benchmarking & monitoring
  • Designed OCR + Claude pipeline for structured content extraction
  • Scaled to 50K+ queries/hour with FastAPI microservices + Guardrails AI
RAG
Weaviate
CLIP
FastAPI
Guardrails
RAGAS
sde_intern.service.ts
SDE Intern
@Zinier
Jan 2025 – Jul 2025
  • Built LangGraph AI agents for anomaly detection → automated scheduler alerts (500+/day)
  • Fine-tuned Llama-3 70B (LoRA) with 10K+ domain samples → +35% accuracy
  • Shipped NL-to-workflow pipeline (FastAPI + React + PostgreSQL, Docker on AWS) → 200+ workflows/hr
LangGraph
Langfuse
LoRA
Llama-3
FastAPI
React
AWS
ai_intern.service.ts
AI Engineering Intern
@PDH
Oct 2024 – Jan 2025
  • Built hybrid retrieval (dense + BM25) with Pinecone + Cohere reranker → 95% accuracy
  • Optimized infra with Redis caching + async batching → -20% cost, +15% speed
  • Created custom eval framework with TF-IDF scoring + A/B testing
Pinecone
Cohere
Redis
FastAPI
ZenGuard
# Fun fact: My git commit messages are better than my dating profile
# Also: I debug in production because I like to live dangerously (just kidding)
Featured Projects
$ ls -la ~/projects/

Traffic Insight Dashboard

A dashboard that shows live jams, digs through past ones, and even explains why your evening commute always feels cursed, thanks to a graph powered AI that actually understands traffic.

Real-time Traffic
Analytics
Incident Tracking
Graph RAG
Congestion Insights

Web Search Agent

Built an agent that researches Wikipedia and serves results via an API. Because Googling is easy but reading isn’t.

Research Agent
API Design
Web Scraping
LLM Tools

Finance MCP server

Created a financial data agent that fetches and analyzes live stock/crypto data, Because Bloomberg terminals shouldn’t have all the fun

Finance API
MCP
Data Agents
Real-time Retrieval

Agentic AI Coach

Prototyped a multi-agent system where specialized AI coaches advise founders across UX, marketing, and decision-making, orchestrated through a shared memory and tool layer.

Multi-Agent Systems
LangChain
AI Coaching
Orchestration

Internal-Helpdesk-Portal

Built an AI-driven internal helpdesk with ticket routing, agent-assist replies, and misuse detection. A modular FastAPI + MongoDB system where LLMs act as both triage and watchdog.

FastAPI
MongoDB
WebSockets
AI Routing
RAG

Internal-Helpdesk-Portal

Built an AI-driven internal helpdesk with ticket routing, agent-assist replies, and misuse detection. Because even support desks need support ,here LLMs act as both triage and watchdog.

FastAPI
MongoDB
WebSockets
AI Routing
RAG

civic-chain

Developed a petition system with web 3 and AI agents with verifiable, append-only logs inspired by blockchain design, ensuring civic data is tamper-resistant.

Blockchain-inspired
Governance Tech
Audit Logs
Web App
AI Agents
OpenAI

hippocamus-extension

Built a Chrome extension that lets users save, tag, and semantically search their online finds with an AI-powered memory layer. Basically a smarter Ctrl+F that actually remembers context.

Chrome Extension
Semantic Search
RAG
PineCone
FastAPI
/* These projects are like my children - they keep me up at night and have bugs I can't explain */
Education
$ cat /var/log/education.log
Misogi AI
AI Engineering
Focused on building production-grade AI systems: LLMs, RAG, Agents, and MLOps for real-world deployments.
LLMs
RAG
LangGraph
AI Agents
MLOps
Prompt Engineering
Computer Vision
NLP
FastAPI
Vector DBs

/* Shipping AI to prod: accuracy ↑, latency ↓, cost ↘ */

Birla Institute of Technology and Science, Pilani
B. Sc. Computer Science
Where I learned that semicolons are not optional and indentation matters
Computer Science
Data Structures
Web Development
Operating Systems
Database Management Systems

// Graduated with a degree and caffeine addiction

Continuous Learning
Because in tech, if you're not learning, you're debugging outdated code.
while(alive) {
learn();
code();
debug();
coffee++;
}
Fun Corner
$ echo "Because life isn't all about semicolons"
Life.interests()
while (energy > 0) {
swimming();
exploreMountains();
studyAI();
roadTrip();
}
When not fixing bugs, you'll find me:
  • Swimming (because floating is better than null pointer exceptions)
  • Road tripping (because life is a journey, not a destination)
  • Hiking (because mountains have better views than code)
Dev Jokes.js
// Why do programmers prefer dark mode?
Because light attracts bugs! 🪲
try {
coffee.drink();
code.write();
} catch (CoffeeEmptyError) {
orderMore();
}
My code doesn't have bugs,
it just develops random unexpected features! 🎯
Contact
$ ssh user@ansuman
// Initialize contact methods
Email
ansuman.ai@proton.me
LinkedIn
www.linkedin.com/in/ansuman-shukla
X
x.com/ansuman_bin
// Quick Response Promise

async def response_time(has_coffee):
if has_coffee:
return "ASAP! (Running on asyncio and espresso ☕)"
return "Soon (after a quick coffee coroutine)"
About
Passionate AI Engineer crafting elegant solutions to complex problems.
Contact
Location
Based in Bengaluru, India

© 2025 Ansuman. All rights reserved.

Made with ❤️ in India