Ajay Bharathi
I forge AI like modern Prometheus — turning data into fire, and code into creation. From agents that restructure chaos into clarity to models that learn and evolve,
I build systems where myth meets math.
About Me
AI Developer & Research-Oriented Engineer passionate about building intelligent, scalable systems at the intersection of Machine Learning, Deep Learning, and real-world applications. I specialize in designing AI-driven solutions across domains including Computer Vision, NLP, and autonomous systems, with a strong focus on turning research ideas into production-ready products.
With hands-on experience in developing end-to-end AI pipelines—from data processing and model training to deployment using cloud and microservices—I have worked on impactful projects such as AI-powered career intelligence platforms, satellite-driven agricultural systems, and physics-informed lunar terrain reconstruction under ISRO guidance. My work spans across advanced areas like Retrieval-Augmented Generation (RAG), AI agents, computer vision, and hybrid ML architectures.
Beyond building models, I take a system-level approach by integrating AI with backend engineering, cloud infrastructure, and real-time data pipelines to create scalable and efficient solutions. I actively explore generative AI, transformers, and intelligent automation, aiming to bridge research with real-world impact and contribute to solving complex challenges across domains like space, cybersecurity, agriculture, and enterprise systems.
Project Overview
// 8 active transmissions — hover to pause, click to explore
Mission Chronicle
Building a MAML-based Universal Dynamic Adaptive Predictive Model and AI-driven multi-tenant dynamic integration platform. Meta-learning framework for continual adaptation across diverse task distributions.
1st place nationally for the NTRO Cybersecurity problem statement. AEGIS-Ω Hybrid IDS with BiLSTM anomaly detection identifying 13+ URL attack types at 97% accuracy across real-world traffic.
Career Intelligence platform with resume AI, psychometrics, RAG roadmap generation, and Gemini mock interview workflows — shortlisted at Google Cloud Gen AI Exchange Hackathon 2025.
Completed Advanced Retrieval Augmented Generation certification covering chunking strategies, vector retrieval, re-ranking, and production RAG evaluation pipelines.
Physics-AI hybrid pipeline combining Shape-from-Shading photometric modelling with CNN residual learning for high-resolution lunar DEM generation from mono imagery. ±0.3m RMSE accuracy.
VS Code extension for AI-assisted Python project restructuring selected at IBM TechXchange 2025. Auto-generates docs, boilerplate, and refactors imports using Google Gemini agents.
Satellite-driven agricultural intelligence platform integrating geospatial data, weather APIs, and soil data for crop recommendations and sustainability MRV insights.
Built production-grade 4-layer IDS: Signature → Autoencoder → BiLSTM → Ensemble. Real-time PCAP analysis with async processing, 96.8% accuracy, 1334 pkt/s CPU throughput. Led to SIH win.
Column-aware OCR pipeline using EasyOCR + YOLO layout detection, LLM post-processing for error correction, and ElevenLabs TTS for automated audio accessibility output.
Fine-tuned YOLOv8 Nano for small and distant face detection in CCTV / crowd scenarios. Trained on WIDER FACE dataset, exportable to ONNX for edge deployment. mAP@50 ~0.64.
Certified in Disaster Risk Monitoring Using Satellite Imagery — deep learning on remote sensing data for humanitarian and climate applications.
Founded Cryptarchs — an AI-first development collective focused on building production-grade intelligent systems across cybersecurity, computer vision, space AI, and developer tooling. All major projects built under this banner.
Introduction to Generative AI — LLM foundations, prompt engineering, model fine-tuning, and responsible AI principles on Google Cloud infrastructure.
Secure load-forecasting pipeline using CKKS homomorphic encryption and differential privacy — encrypted time-series inference without exposing plaintext energy data.
Built REST APIs and backend services for hostel management platform. Automated fee tracking and document workflows for 1500+ records. Integrated RAG-based document search, reducing manual effort by 40%.
Oracle Foundations Associate: Blockchain & Cloud Computing — distributed ledger fundamentals, smart contracts, and cloud infrastructure design principles.
Built autonomous rover from scratch with tri-mode control (voice, mobile, signal), face recognition, object tracking, line following, and obstacle avoidance — total hardware cost under ₹12,000.
Sacred Inscriptions
// Verified certifications — hover to pause, click to explore