I build machine learning systems, LLM applications, backend services, and developer-focused software. My work focuses on turning research ideas into reliable, testable implementations.
lakshmi@sagar:~$ whoami
Lakshmi Sagar Seshadri
Focus : AI/ML, LLMs, RAG, Graph AI, Backend Systems
Engineering: Python, TypeScript, Java, C, C#
Stack : PyTorch, TensorFlow, FastAPI, ASP.NET Core, React, Angular
Data : PostgreSQL, pgvector, Supabase, SQL Server
Infra : Linux, Docker, Git, GitHub Actions, Vercel
| Project | Description | Stack |
|---|---|---|
| SAGE | Knowledge-graph reasoning system for temporal relationships, suppression semantics, provenance, and controlled analytical question answering. | Python, NLP, Knowledge Graphs |
| DNS Sentinel | Real-time DNS tunneling detection platform using behavioral features and an ML ensemble for detecting covert DNS activity. | Python, FastAPI, XGBoost, Random Forest, PostgreSQL |
| Research Assistant | Full-stack research workflow covering literature search, summarization, citation management, gap detection, and review generation. | Python, FastAPI, React, TypeScript |
| Mood Journal | Full-stack application for daily mood logging, emotional pattern tracking, history, and analytics. | Angular, ASP.NET Core, SQL |
| CropGuard | Crop health platform for disease detection, condition monitoring, analytics, and farmer-focused recommendations. | React, Node.js, Express, MongoDB |
| SoilTwin | Digital-twin platform for soil analysis combining web interfaces, analytics, data processing, and machine learning. | Next.js, TypeScript, Python |
| LFAN | Lightweight feature-attention network for efficient ×4 single-image super-resolution on DIV2K. | Python, Deep Learning, Computer Vision |
| AssumptionX | AI-powered application for identifying and challenging hidden assumptions in ideas, plans, and decisions. | TypeScript, Next.js, React, Tailwind CSS |
| RagaAI Catalyst | LLM project platform covering dataset management, evaluation, tracing, prompt management, synthetic data generation, and guardrails. | Python, LLMs, Evaluation, Tracing |
- Retrieval-Augmented Generation and evaluation
- Graph-based and temporal reasoning
- LLM systems and local inference
- Model evaluation and reliability
- Fraud detection and federated learning
- Multi-agent AI systems
[ BUILDING ] Reliable AI and LLM systems
[ BUILDING ] Retrieval and verification pipelines
[ BUILDING ] Graph reasoning systems
[ RESEARCH ] Model evaluation and reliability
[ LEARNING ] Deep learning architectures and attention-based models
lakshmi@sagar:~$ exit