Ahamed H

Building AI for real-world solutions

Computer Vision · RAG · LLM Systems · Deep Learning

Open to AI/ML opportunities

// exploring AI techniques & building enterprise-grade solutions

Building something interesting? If you have a project I could contribute to, let's talk — mail me at ahamedh2003@gmail.com.

ABOUT

I'm an AI Engineer and 2026 Computer Science graduate working across computer vision, retrieval systems, and LLM-powered applications. I've built and deployed AI systems involving model training, hybrid RAG pipelines, REST APIs, Docker, and Google Cloud Run. During DeLTA 2026 at IIIT Allahabad's Computer Vision & Biometrics Lab, I completed intensive deep learning training and worked on a mentored project exploring self-supervised learning under low-label regimes.

Ahamed H portrait
2801SafeGuard training images
0.702SafeGuard mAP50
0.877SafeGuard precision
1.0DocMind RAGAS faithfulness
0.845DocMind overall RAGAS score
10 / 20 / 50%SSL labeled-data regimes tested
01WORK

Selected Projects

Projects across computer vision, RAG, and LLM-powered applications — from dataset preparation and model training to API design and cloud deployment.

Computer Vision / Object Detection

SafeGuard

Real-time PPE detection system built by fine-tuning YOLOv8n on a 2,801-image dataset containing 10 PPE classes.

mAP500.702
Precision0.877
Dataset2,801 images
Classes10
PythonYOLOv8nOpenCVStreamlitDockerGoogle Cloud Run
RAG / Information Retrieval

DocMind

Hybrid retrieval-augmented generation system combining dense and sparse retrieval with fusion, reranking, and evaluation.

RAGAS Faithfulness1.0
Overall Score0.845
Query
Query Reformulation
FAISS + BM25
Reciprocal Rank Fusion
Cross-Encoder Reranking
Gemini 2.0 Flash
Grounded Answer
PythonFAISSBM25Sentence-TransformersGemini 2.0 FlashRAGASFastAPIStreamlit
LLM Application / Retrieval

Podcast Intelligence

Podcast question-answering system that retrieves YouTube transcripts and produces grounded answers with timestamp citations.

YouTube URL
Transcript Extraction
Sliding Window Chunking
Keyword Retrieval
Gemini 2.5 Flash
Timestamp-grounded Answer
PythonFastAPIGemini 2.5 FlashYouTube Transcript APIReact
02EXPERIENCE

Experience

2026

Computer Vision & Biometrics Lab (CVBL), IIIT Allahabad

Deep Learning Intern

Jun 2026

DeLTA 2026 — Deep Learning Technology and Applications

Completed intensive deep learning training and a mentored project on self-supervised learning under low-label regimes.

  • Training covered CNNs, GANs, RNNs, Transformers, and LLMs
  • Explored self-supervised learning to reduce dependency on labeled training data
  • Evaluated model performance under 10%, 20%, and 50% label-availability settings
  • Work involved literature review, experimental configuration, training, and results analysis
10%%
10%
20%%
20%
50%%
50%
DEEP LEARNING INTERNCVBL · IIIT ALLAHABADDELTA 2026JUN 2026ONLINE
Program details
2025

ShadowFox

Python Development Intern

Aug 2025

Built Python automation scripts and REST API integrations. Debugged data processing pipelines and improved API response handling.

03ARSENAL

Technical Arsenal

AI / ML

Machine LearningDeep LearningDeLTA 2026Computer VisionSafeGuardNLPSelf-Supervised LearningDeLTA 2026RAGDocMindLLMsDocMind, Podcast IntelligenceTransformers

Retrieval / LLM

FAISSDocMindBM25DocMindSentence-TransformersDocMindRAGASDocMindGemini APIDocMind, Podcast IntelligenceLangChain

Vision

YOLOv8SafeGuardOpenCVSafeGuardRoboflowSafeGuard

Backend

FastAPIDocMind, Podcast IntelligenceREST APIsDockerSafeGuard

Cloud

Google Cloud RunSafeGuardGCPSafeGuard

Languages

PythonSQLJavaScript

Tools

GitJupyterGoogle ColabStreamlitSafeGuard, DocMindVS Code
05CREDENTIALS

Workshops & Certifications

Generative AI & Large Language Models Workshop

IIIT Nagpur & NVIDIA Deep Learning Institute

March 2026

Machine Intelligence and Vision Algorithms (MIVA-2024)

NIT Kurukshetra

September 2024

04CONTACT

Get in Touch

Scrolled this far?

If you're working on an interesting AI problem, let's talk.

View ResumeLinkedIn