CV
Curriculum Vitae of Jha Saket Sunil
Contact Information
| Name | Jha Saket Sunil |
| Professional Title | Student |
| saketjha0324@gmail.com | |
| Phone | +91 9370456334 |
Professional Summary
This is Saket. I am currently studying at the National Institute of Technology Karnataka (NITK), Mangalore, India. I was born and brought up in Mumbai, Maharashtra, where I spent most of my early years. I am deeply interested in AI & Maths. I enjoy exploring how AI can be used to solve real-world problems, especially in areas like education technology, automation, and large language model applications. I like experimenting with new ideas, building projects, and documenting my learning along the way.
Experience
-
May 2025 - July 2025 Remote-Hybrid
AI Engineer Intern
Kalppo
- Developed a scalable pipeline using Google Cloud Vision, structured content generation, and fuzzy sequence matching to automate student workbook evaluation, reducing manual assessment time by 80%.
- Containerized a FastAPI-based REST service for input-driven STEM question and SMILES code generation, featuring a LaTeX-to-image rendering pipeline and Supabase integration for media storage.
- Collaborated on building an AI-driven Microservice for automated question extraction from PDFs using Mistral OCR, LangChain, and Gemini for structured parsing, with PostgreSQL for metadata storage.
Education
-
2023 - 2027 Karnataka, India
B.Tech
National Institute of Technology Karnataka, Surathkal
Civil Engineering, Minor in Information Technology
-
2021 - 2023 Dadar, Mumbai
Junior College
Kirti M Doongursee College of Arts, Science and Commerce
12th Science
Awards
-
June Winner, Kaggle x Skill Assessment Machine Learning Competition– Rank 1/1846
Kaggle
Skills
Languages
Interests
Projects
-
FileForge
A distributed file storage system enabling secure file and folder management with scalable backend architecture.
- Built a distributed file storage system supporting nested folders, uploads, ZIP extraction, and favorites.
- Implemented secure authentication and session management using JWT and OAuth2.
- Integrated MinIO object storage with PostgreSQL metadata storage and optimized indexing.
- Achieved ~40% faster file metadata queries with optimized database design.
- Designed modular FastAPI routes and containerized deployment using Docker.
-
ATLAS – Adaptive Traffic Light Allocation System
A computer vision based adaptive traffic signal system that dynamically allocates signal timings using real-time vehicle detection and tracking.
- Developed real-time vehicle detection and tracking using YOLOv8 and ByteTrack.
- Implemented Region Masking and frame-based confirmation filtering for accurate vehicle counting.
- Designed DBWSA algorithm to dynamically allocate green signal time based on vehicle density and class.
- Achieved 25% average and 50% peak efficiency improvements over fixed traffic signals across 500+ simulations.
- Built a flexible pipeline supporting multi-source video input, YAML configuration, and Pydantic validation.
-
FilmFusion
A hybrid movie recommendation system combining semantic embeddings with content-based filtering.
- Implemented hybrid recommendation using Sentence Transformers and semantic similarity.
- Used FAISS vector indexing for efficient embedding search and retrieval.
- Enabled dynamic filtering by genre, language, popularity, runtime, and release year.
- Built an interactive web interface using Streamlit.
Extracurricular Activities
- Web Club NITK — Member
• Presented and explained Mamba SSM and its advantages in efficient sequence modeling.
• Discussed modern alternatives to transformer-based architectures.
- IE NITK (Institution of Engineers) — Member
• Assisted in developing a YOLOv5-based fire-fighting robot.
• Achieved 86% mAP@0.95 object detection performance.