Project Intern

Indian Statistical Institute (ISI), Electronics and Communication Sciences Unit (ECSU)

June 2024 - Present | Kolkata, India (Onsite)

Supervisor: Dr. Swagatam Das

• Worked on simplified Medical Report Generation (MRG) using prompt engineering (Chained Prompting, In-Context Learning), fine-tuned LLaMa models, and integrated Gemini-1.5-Flash API.
• Proposed CPMK-E (Chained Prompting for Medical Report Keyword Extraction) for LLM fairness in medical settings (Accepted at AAAI 2025 Student Abstract).
• Developed a detailed image captioning system using BLIP, CerberusDet (fine-tuned YOLOv8), fine-tuned LLaMa 3.1/2, Mistral 7B, Phi-2, and Gemini-1.5-Flash API. Evaluated with BLEU, ROUGE, BERTScore, Repetition Rate.

Research Internship

Defence Research and Development Organisation (DRDO), DEBEL Lab

March 2024 - April 2024 | Bengaluru, India (Onsite)

• Developed a comprehensive dataset using IMU sensor data (Xsens) capturing walking patterns of 20 subjects.
• Applied computer vision and machine learning for motion analysis.
• Researched human gait pattern analysis using ensemble learning for real-time prediction systems for lower limb prosthetics.

Machine Learning Engineer Intern

Geogo Techsolutions

October 2023 - Present | Kolkata, India (Hybrid)

• Built an end-to-end face comparison and similarity search solution (98% accuracy vs. Amazon Rekognition) using state-of-the-art CV models for employee authorization and verification.
• Developed a custom vector database with Bruteforce and PCA indexing for efficient large-scale image embedding searches.
• Engineered an Automatic Speech Recognition (ASR) model (Transformer architecture) for Hindi-English mixed data, achieving 34% WER with innovative audio preprocessing (STFT, Conv1D). Deployed for insurance call transcription.