Saksham Kumar

About

Computer Science undergraduate with hands-on experience building full-stack applications and Generative AI systems using FastAPI, Next.js, LangGraph, RAG, and LLMs. Published researcher on a hybrid ML + LLM framework for student career guidance.

Education

• B.Tech
  Oriental Institute Of science and Technology
  Computer Science · 2023 – 2027

Skills

Tools / apps / platforms

Projects

• AtlasPath — Graph-Based Global Travel Route Optimizer ↗
  C++

  Developed a C++ travel planner using Dijkstra's Algorithm across 200+ cities, supporting multi-modal transport with Google Maps visualization via automated HTML reports. Optimized shortest-path computation using graph data structures for efficient route planning based on distance, travel time, and transportation mode.

• Cognitive SOC Framework — Multi-Agent AI Security System ↗
  Random Forest, Isolation Forest, SHAP, LIME, FastAPI, Docker

  Engineered a 14-agent cybersecurity architecture for real-time threat detection and automated incident response with confidence scoring and risk prioritization. Built an explainable AI pipeline served via FastAPI and containerized with Docker for scalable, real-time data ingestion and transparent threat prediction.

• Meeting Intelligence Agent ↗
  FastAPI, Next.js, LangGraph, OpenRouter

  Built a full-stack meeting-intelligence agent converting raw transcripts into structured summaries, decisions, and action items via a 5-stage Extraction–Validation–Correction workflow with live SSE progress streaming. Engineered a validation layer and added one-click PDF report export, SQLite-backed searchable history, and a pytest suite with mocked LLM calls.

• Agentic AI Learning & Assessment Platform ↗
  FastAPI, Next.js, LangGraph, SQLite

  Built a scalable full-stack AI learning platform featuring persistent memory, PDF chat, coding analytics, and automated mock assessments across 5 categories. Developed agentic AI workflows for coding-profile analysis, skill-gap detection, and personalized DSA interview recommendations using LLM-based reasoning and RAG pipelines.

Courses & certifications

• Agentic AI for Developers: Concepts and Applications for Enterprises
• SQL for Data Analysis

🏆 Achievements & awards

• Competitive Programming

  Solved 600+ LeetCode problems covering Data Structures, Algorithms, and Dynamic Programming.

• First Runner-Up, Pareto Time'25, Kirori Mal College (DU) · 2025

  Ranked in the top 50 of 2,000+ participants; presented data-driven policy solutions to domain experts.

• Published research: "Reducing Decision Uncertainty in AI-Based Student Career Guidance Using a Hybrid ML & LLM Framework" · 2026

  Published in IJSRTM.

📚 Publications

• Reducing Decision Uncertainty in AI-Based Student Career Guidance Using a Hybrid ML & LLM Framework · 2026