Data Engineer turning
raw data into AI products.
I'm Saurav — I build retrieval pipelines, multi-agent systems, and the data foundations that make machine learning actually ship.
Saurav
chsaurav · @charlesaurav13
GitHub at a glance
Repos
AI / Data
Years coding
Followers
About
I'm obsessed with the messy, beautiful work of making AI useful.
Data Science is my favorite topic. I love sitting at the seam between pipelines and models — moving data from somewhere ugly into something a model can reason over. Lately I've been building RAG systems, multi-agent harnesses, and trading bots powered by LLMs.
Experience
LinkedInBusiness Technology Solutions Associate
CurrentZS Associates · Full-time
Jun 2024 — Present• Gurugram, India · On-site
- Cut API-driven data ingestion latency from hours to under 2 minutes via parallel processing, batching, and workflow tuning.
- Migrated legacy Azkaban pipelines to modern orchestration; designed Airflow pipelines powering real-time campaign analytics for major brand launches.
- Built curated data models, KPI layers, and analytics-ready datasets; optimized distributed processing to reduce compute cost.
- Exploring ML, NLP, and GenAI — hands-on with RAG architectures, vector DBs, and LLM-powered apps.
Language mix
- Python48%
- Jupyter22%
- JavaScript18%
- TypeScript / Other12%
Currently
- Building RAG pipelines and multi-agent systems on top of vector DBs and LLM gateways.
- Shipping production Airflow pipelines for real-time campaign analytics at ZS Associates.
- Exploring LLM-powered trading bots— strategy research, backtesting, and execution.
2+
Years building
10+
Pipelines shipped
∞
Curiosity
Tech stack
Tools I reach for, grouped by what they do
Multi-agent AI
Orchestrating LLM agents that plan, call tools, and self-correct.
RAG & search
Vector retrieval, chunking, and grounding LLMs on real documents.
Applied ML
Computer vision, trading models, and end-to-end data pipelines.
Featured work
Selected projects
AutoHedge
Python
AI-powered crypto trading analysis using DeepSeek + Swarms multi-agent framework. Technical indicators, risk management, and automated trade signals for BTC, SOL, ETH.
Payer_Rag_ADS
Jupyter
RAG pipeline extracting 12 prior-authorization parameters from PsO payer policy PDFs. Production document intelligence.
RAG-System
Python
End-to-end retrieval-augmented generation pipeline — chunking, embeddings, vector search, and grounded answers.
hive
Python
Multi-agent harness for production AI. Orchestration patterns for reliable agentic workflows.
autohedge-trading-bot
Python
DeepSeek RL agent with OpenClaw memory context — a live trading bot iteration on the AutoHedge research.
Driver_Drowsiness_Detection
Jupyter
Computer-vision model predicting driver drowsiness from eye state — applied deep learning for safety.
Let's build
Have a data or AI problem worth solving?
I'm open to collaborations, full-time roles, and interesting AI/Data projects. Reach out — the inbox is always open.
GitHub
@charlesaurav13
in/saurav7109
hello@saurav.dev