Hello, I'm

Sakhinala Sanjay Bhargav

AI/ML Engineer

Building and Deploying Real ML Systems

Open to AI/ML opportunities

AI/ML engineer specializing in reinforcement learning and NLP. Built a production multilingual RAG pipeline for 10+ Indian languages, and an uncertainty-aware RL system for electromagnetic structure optimization that cut simulation costs by 72%. Authored 3 IEEE papers (1 published, 2 in press) and filed a patent in antenna design for 6G.

View My Work Get In Touch Download Resume
Sakhinala Sanjay Bhargav
// About

About Me

I build and ship production ML systems — not just notebooks. My work includes a multilingual RAG pipeline supporting 10+ Indian languages, deployed as a LangGraph agentic RAG system with reflexion and versioned to v2.3.0, and an uncertainty-aware reinforcement learning system that cut antenna simulation cost by 72% compared to genetic algorithms and particle swarm methods.

My research produced 3 IEEE conference papers — one published, two in press — presented or accepted at IIT Bhubaneswar, IIT Indore, and BVRIT Hyderabad, plus a filed Indian patent. I'm based in Hyderabad, Telangana, and I'm looking for an entry-level AI/ML or software engineering role.

// Impact
0 %

Cost Reduction

simulation cost reduction vs. GA/PSO baselines

0 +

Indian Languages

supported by the IndicRAG QA system

0 %

Model Accuracy

on EuroSAT Land Use Classification

0

IEEE Papers + Patent

1 published, 2 in press; patent filed in 6G antenna design

// Projects

Selected Work

Uncertainty-Aware RL for EM Structure Optimization
Research & Publication
Accepted — WAMS 2026

Uncertainty-Aware RL for EM Structure Optimization

A reinforcement learning system that slashes simulation costs for 6G antenna design by 72%. Combines Soft Actor-Critic with a LightGBM surrogate model and uncertainty quantification to minimize expensive CST electromagnetic simulations while maintaining design quality.

PyTorch Stable-Baselines3 LightGBM Optuna CST API
Read the Write-up
Multi-Agent Pipeline
Open Source

AI Data Scientist Platform

A multi-agent data science pipeline powered by CrewAI. Seven specialist agents load, clean, engineer features, run statistics, train models, visualize results, and write an executive report — end to end from a single dataset.

CrewAI Python scikit-learn pandas Streamlit Docker
GitHub
Observability Platform
Production

AgentOps

A mini-LangSmith for AI agent observability, debugging, and evaluation. Records agent executions as interactive graphs, scores faithfulness and hallucination, and tracks cost and latency across your entire agent fleet.

FastAPI Next.js TypeScript PostgreSQL React Flow Python SDK
GitHub Live Demo
IndicRAG - Multilingual document QA system
Multilingual RAG
Production

IndicRAG

A multilingual document QA system supporting 10+ Indic languages. Combines hybrid dense + sparse retrieval with a LangGraph agentic pipeline that plans, uses tools, retrieves across sources, and critiques its own answers.

LangGraph HuggingFace ChromaDB NLLB-200 Gemini API FastAPI
GitHub

More Projects

CipherChat - Secure encrypted messenger
Open Source

CipherChat

End-to-end encrypted messenger with zero-knowledge architecture. AES-256 + RSA-2048 hybrid cryptography over TCP.

Python Cryptography TCP/IP
Open Source

EuroSAT Land Use Classification

CNN with Grad-CAM explainability for satellite imagery. 89% accuracy across 10 land-use categories.

PyTorch TorchVision Grad-CAM
ResolverLab - DNS resolver benchmarking
Open Source

ResolverLab

DNS resolver benchmarking with statistical rigor. Automated performance analysis for enterprise infrastructure.

Python Network Programming Data Viz
// Publications

Publications & Patents

Patent

🔬

Design of Hexagonal Patch Antenna at 28 GHz

Inventor: Sakhinala Sanjay Bhargav

Application No.: 202541014595 A | Published: March 7, 2025 (India)

Publications

📄

Bandwidth Enhancement of Slotted Hexagonal Patch Antenna for 6G Ultra-Fast Data Transfer and Brain-Computer Interface

Presented at ICMOCE 2025, IIT Bhubaneswar

DOI: 10.1109/ICMOCE64100.2025.11076991 July 17, 2025
📄

Bandwidth Optimization of Slotted Circular Patch Antenna for 6G Ultra-Fast Data Transfer and Brain-Computer Interface Application

Presented at the 16th ICCCNT 2025, IIT Indore

Proceedings submitted to IEEE Xplore — Publication Pending
📄

Uncertainty-Aware Reinforcement Learning System with Blended Surrogate Models for Electromagnetic Structure Optimization

Accepted at WAMS 2026, BVRIT Hyderabad

// Training

Training Programs

Generative AI Training

SkillHive Connect

Aug 2025 – Dec 2025
  • Developed a strong foundation in Generative AI, Machine Learning, Deep Learning, and NLP using Python
  • Worked with Large Language Models (LLMs), Transformer architectures, LangChain, and Retrieval-Augmented Generation (RAG)
  • Performed fine-tuning of LLMs including hands-on exposure to LLaMA models
  • Built real-world GenAI mini-projects such as AI chatbots, semantic search engines, YouTube transcript summarization, and a voice-based medical assistant

AI/ML Training

Google

Jan 2025 – Mar 2025
  • Developed computer vision modules using Python and OpenCV to automate image detection tasks for mobile applications
  • Evaluated model performance and deployed ML pipelines using Google Vertex AI and Model Deployment APIs
  • Integrated ML features into end-to-end app workflows, enabling real-time inference and improved automation efficiency

Cybersecurity Training

Palo Alto Networks

Oct 2024 – Dec 2024
  • Implemented firewall policy optimization to strengthen network access control and reduce alert false positives
  • Analyzed threat logs and worked with SIEM workflows to identify anomalies using intelligent log-correlation systems
  • Gained working knowledge of secure network design, intrusion detection fundamentals, and endpoint protection frameworks

Data Science Applied Training

Altair RapidMiner

Apr 2024 – Jun 2024
  • Built no-code ML pipelines in RapidMiner for data preprocessing, feature engineering, and model evaluation
  • Applied supervised learning algorithms such as SVM and Random Forest for classification and performance benchmarking
  • Performed exploratory data analysis (EDA) and interpreted results to derive actionable insights
// Skills

Technical Skills

🧠

AI & Machine Learning

Deep Learning Computer Vision Reinforcement Learning SAC & PPO Transformers RAG Multilingual NLP Prompt Engineering
⚙️

ML Frameworks

PyTorch TensorFlow scikit-learn Stable-Baselines3 HuggingFace TorchVision OpenCV LightGBM Optuna
🚀

MLOps & Deployment

Docker FastAPI Model Serving Vertex AI REST APIs FAISS Redis MLflow CI/CD Git Google Cloud
📊

Data Engineering

Pandas NumPy SQL Matplotlib Seaborn
🔒

Security & Systems

Python Linux CLI Cryptography RSA & AES Networking TCP/IP DNS Shell Scripting
// Credentials

Certifications & Education

Education

B.Tech in Electronics and Communication Engineering

Bapatla Engineering College, Andhra Pradesh

May 2025

Diploma in Electronics and Communication Engineering

Aditya College of Engineering, Surampalem, AP

Oct 2021

Certifications

✓

Google Data Analytics Professional Certificate

SQL • R Programming • Data Cleaning • Tableau • Business Communication

March 2025
✓

Google Cybersecurity Professional Certificate

Network Security • Python • Linux • Cloud Security • Network Architecture

May 2024
✓

Google Prompting Essentials

Prompt Engineering for GenAI Workflows

May 2025
✓

AI For Everyone — DeepLearning.AI

AI Strategy • ML Fundamentals • Business Use-Cases

February 2025
✓

Google Skill Badges (10+)

TensorFlow • Image Classification • Object Detection • Product Search

2024-2025
✓

NPTEL: Cloud Computing

Engineering Architecture • Virtualization Concepts

2024
✓

NPTEL: Computer Networks & Internet Protocols

TCP/IP • Routing • DNS • Network Performance

2024
// Blog

Blog

IndicRAG architecture workflow
July 16, 2026

IndicRAG: The Jump from Standard RAG to Agentic RAG

How the v2.0.0 rewrite turned IndicRAG's single-pass retriever into a LangGraph agent with tool use and reflexion — and what four releases of hardening after that jump actually fixed.

Read More →
ResolverLab DNS benchmarking suite
February 4, 2026

ResolverLab: High-Performance DNS Benchmarking Tool

A production-grade Python analytics suite for benchmarking DNS services across 50+ providers, measuring block rates, latency, and cache performance with comprehensive reporting.

Read More →
RL antenna optimization pipeline
January 28, 2026

Uncertainty-Aware Reinforcement Learning for 6G Antenna Design

Exploring how uncertainty-aware RL agents can optimize electromagnetic structures, reducing simulation costs by 72% through hybrid surrogate modeling approaches.

Read More →
IndicRAG architecture workflow
December 8, 2025

IndicRAG: Breaking Language Barriers in Document QA

A practical guide to implementing production-ready retrieval-augmented generation pipelines for Indian languages with semantic chunking and cross-encoder reranking.

Read More →
End-to-end encrypted messenger architecture
November 22, 2025

End-to-End Encryption in Python: Building Secure Messaging Systems

Learn how to implement hybrid AES-256 and RSA-2048 cryptography with secure key exchange protocols for zero-knowledge architecture messaging.

Read More →
// Contact

Get In Touch

I'm always interested in new opportunities, collaborations, or just a conversation about AI/ML. Feel free to reach out!

sanjaysakhinala@gmail.com
Hyderabad, Telangana, India