Fahad S.

Key Skills

Software

Other

Top Subject Matter

Dialogue Summarization

LLM Fine-tuning

Machine Learning

Top Data Types

Text

Image

Audio

Top Task Types

Text Summarization

Classification

Text Generation

RLHF

Fine-tuning

Prompt + Response Writing (SFT)

Freelancer Overview

LLM Fine-tuning and Dialogue Summarization Project. Brings 3+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Other. Education includes Master of Technology, IIT Kharagpur (2026) and Intermediate, Islamia College, Darbhanga (2021). AI-training focus includes data types such as Text and labeling workflows including Text Summarization and Classification.

Labeling Experience

LLM Fine-tuning and Dialogue Summarization Project

OtherTextText Summarization

Fine-tuned FLAN-T5 on the DialogSum dataset utilizing the LoRA technique to enhance dialogue summarization performance. Implemented Reinforcement Learning from Human Feedback (RLHF) with PPO to further improve model behavior and reduce toxicity in generated summaries. Utilized a hate speech classifier for reward modeling and achieved a substantial reduction in toxicity. • Created sophisticated dialogue summaries using advanced LLM training techniques • Labeled and generated high-quality data for fine-tuning on text summarization tasks • Applied RLHF and PPO methods for safer output generation • Employed third-party evaluation models to assess and refine output

2025 - 2025

Social Care Data Labeling & Risk Classification for LLMs

OtherTextClassification

Preprocessed and engineered social care features to create a risk classification system using LLMs and BERT models. Fine-tuned BERT-Base-Uncased Model for classifying census tracts into risk categories based on SDOH data. Optimized dataset creation and feature selection to improve downstream model performance for policy-making. • Created annotated datasets for BERT-based risk classification • Labeled social care data with risk categories for supervised training • Evaluated model outputs at each stage to ensure classification accuracy • Performed feature engineering supporting data labeling pipelines

2024 - 2024

Education

Work History