U. Jaswant Charith
Jaswant Charith

Currently

Trainee / Intern

ISRO

AI & ML Student
· Aspiring Robotics Engineer

I'm Jaswant Charith, an AI & ML undergrad at REVA University with a pull toward robotics. I'm currently a Trainee/Intern at ISRO and a Founder's Fellow at CropNow, having previously served as its Founder's Office Lead. I build things with computer vision, embedded hardware, and language models along the way.

Reva→BTech CSE (AI & ML), 2023–2027 — where the embedded-systems and IoT work started.ISRO→Trainee / Intern — the current chapter.CropNow→Founder's Fellow

#01 Work

Selected Projects

A few things I've built across computer vision, embedded systems, and language models.

#01

DIVINE Answer

A retrieval-augmented generation app that surfaces guidance from the Quran, Bible, and Bhagavad Gita side by side.

  • Vector database of verses across three scriptures, built on Supabase
  • Cosine-similarity retrieval of the top 3 most relevant verses per query
  • LLM synthesis layer producing a comparative, unified explanation
  • Full stack — frontend, backend, and NLP pipeline — on Supabase Edge Functions

Cosine similarity 0.9974 · MRR 0.5000 · NDCG@5 0.6509

PythonRAGSupabaseOpenAI Embeddings
#02

PARAMETER — Animal Detection & Avoidance

A computer-vision relay system that detects animals in real time and triggers a controlled, non-lethal response.

  • CNN-based detection model deployed on Raspberry Pi for real-time inference
  • Camera input wired to relay-based electrical hardware
  • Logged activity patterns — type and time — for further analysis
  • Tuned for energy efficiency to reduce electricity cost
OpenCVPyTorchRaspberry PiEmbedded SystemsIoT
#03

SANKETHA — Secure LoRa Communication Device

A long-range, low-power messaging device that works completely offline.

  • Reliable long-range data transmission with no internet infrastructure
  • Secure, efficient pipeline for transmitting data over distance
  • Hardware integration tuned for stable, continuous device-to-device link
  • Validated through real-world field testing
ESP32LoRaEmbedded Systems
#04

Smart Attendance System

Facial-recognition attendance tracking that removes manual roll calls from the classroom.

  • Real-time face detection and recognition pipeline
  • Automatic attendance marking with no manual intervention
  • Tuned for accuracy and reliability in live classroom conditions
PythonOpenCVComputer Vision

Tools & Technologies

PythonC++PyTorchTensorFlowOpenCVROSDjangoPostgreSQLAWSAzureESP32