About Me

Who Am I?

Hi I'm Tathagat Sathe, a Bachelor of Technology graduate in Aerospace Engineering from the Indian Institute of Technology Kanpur. I have extensive professional experience building secure, high-throughput backend architectures in the fintech domain, specializing in core investment and personal finance systems serving millions of users.

My expertise bridges the gap between scalable distributed backend engineering and production AI infrastructure. I design and build secure APIs, microservices, and serverless applications on AWS, while integrating robust caching, streaming, and authentication flows.

In the AI space, I specialize in building custom agentic workflows using LangGraph, local Retrieval-Augmented Generation (RAG) pipelines, and optimized deep learning inference runtimes. I focus on deploying end-to-end intelligent systems, ranging from custom PyTorch serving engines with KV-cache to streaming feature aggregation pipelines for real-time model scoring.

Agentic Systems

Backend Systems

Cloud Architecture

Data Infrastructure

My Specialty

My Skills

Python & C++

95%

Javascript & TypeScript

90%

SQL & Databases (PostgreSQL, Redis)

90%

PyTorch & Deep Learning

90%

LangGraph & Agentic Workflows

85%

AWS & Cloud Native

85%

Kafka & Spark

75%

Docker & Kubernetes

80%
Education

Education

Coursework: Autonomous Navigation, Introduction to Economics, Linear Algebra, Partial Differential Equation, Quantum Physics, Optimal Space Flight Control

Projects:
Visual Odometry and SLAM (Simultaneous Localization & Mapping)

Developed and implemented a Simultaneous Localization and Mapping (SLAM) algorithm on an autonomous robotics platform. The algorithm allowed the robot to navigate around an unknown environment without any prior knowledge, using an Extended Kalman filter to fuse wheel encoder odometry and monocular visual odometry data.

Experience

Work Experience

Senior Software Development Engineer - Backend at Freecharge (DBAT) Sep 2024 - current

  • Led multiple end-to-end backend projects, from architecture to deployment, ensuring scalability and security.
  • Designed and implemented secure APIs, database structures, and encryption logic to safeguard sensitive financial data and meet compliance standards.
  • Integrated a secure Single Sign-On (SSO) mechanism across investment products, improving authentication and user experience while maintaining regulatory compliance.
  • Worked closely with product managers, designers, QA, and business teams to align technical solutions with business and regulatory needs.
  • Optimized backend processes and data flows, which improved performance and enhanced the overall user experience for millions of customers.
  • Mentored junior developers and provided technical leadership to drive the delivery of reliable, high-quality solutions.

Assistant Manager at Genpact Nov 2020 - Sep 2024

  • Designed and implemented an automation system, reducing 80% of manual work, resulting in significant cost savings and increased productivity for the company.
  • Designed and implemented CI/CD pipelines to streamline the deployment process, including automated testing and integration.
  • Revamped and modernized legacy codebases, resulting in significant operating cost reductions and improved functionality, leading to a 500% increase in user base.
  • Contributed to all phases of the systems development lifecycle, from requirements gathering to production releases, collaborating with cross-functional teams to ensure the successful delivery of high-quality products.
  • Received three company-wide recognitions, including awards for innovation, speed to outcome, and teamwork, for spearheading the development of a workflow automation system and web applications.
My Work

Recent Work

Maestro | Multi-Agent Research Engine

A multi-agent research workflow engine built with LangGraph, Anthropic Claude, and FastAPI. Orchestrates supervisor, researcher, writer, and critic agents with SSE streaming.

NexusAI | Stock Analysis Copilot

A self-hosted financial research copilot executing custom LangGraph workflows. Features local RAG (ChromaDB + sentence-transformers), Redis caching, and a React dashboard.

FlashInfer | PyTorch LLM Inference Engine

A custom, lightweight LLM inference engine built from scratch in PyTorch. Implements KV-caching, RMSNorm, RoPE, SwiGLU, and Top-K/Top-P sampling with streaming FastAPI endpoints.

CogniPost | Autonomous AI Content Pipeline

An automated daily Twitter/X posting bot. Features Google & X trend discovery, LLM-based fact generation, and multi-stage scoring/guardrail criteria.

Credit Card Fraud Detection | Streaming MLOps

Real-time streaming fraud detection pipeline on AWS. Integrates SageMaker Feature Store, Amazon Kinesis streams, Lambda functions, Spark, and XGBoost.

Hate Speech Detection | Modular NLP

Modular end-to-end NLP classification pipeline using Scikit-learn and PyTorch. Deploys a Streamlit application packaged inside a Docker container.

Conversational Document Query RAG

An intuitive PDF document query application using OpenAI's LLM and Streamlit, enabling users to upload documents and engage in natural language QA.

LLaMA-2 Financial Finetuning

Finetuned LLaMA-2 model on a proprietary financial dataset using PEFT (LoRA) methods to efficiently adapt the LLM for domain-specific advisory tasks.

Certifications

Here are some of my Certifications

IBM Machine Learning

Exploratory data Analysis, Supervised Learning, Unsupervised Learning, Deep Learning and Reinforcement Learning

Deep Learning Specialization

Neural Networks, Hyperparameter Tuning, Regularization, Optimization, Structuring Machine Learning Projects, CNN, Sequence Models

Advanced SQL

JOINs and UNIONs, Anaytic Functions, Nested and Repeated Data, Writing Efficient Queries

Advanced BigQuery

Partitioning, Clustering, Nested Fields, Data Transfer Service, Managing Permissions and Security, Accessing from External Applications

Geospatial Analysis

Coordinate Reference Systems, Interactive Maps, Manipulating Geospatial Data, Proximity Analysis

Mathematics for Machine Learning

Linear Algebra, Vectors, Matrices, Mappings, Eigenvectors

Get in Touch

Contact

Aurangabad, Maharashtra, India