Sanya Ahmed

AI Engineer specializing in LLMs, Agentic AI, and NLP, with hands-on experience building end-to-end intelligent applications. Skilled in fine-tuning transformer models, designing RAG pipelines, and developing multi-agent workflows using PyTorch, Hugging Face, LangChain, and LangGraph. Built and deployed real-time AI systems spanning multimodal inference, conversational interfaces, and agentic tooling — from a personality kiosk to production-grade applications.

Proven ability to move fast, ship independently, and deliver production-ready AI solutions beyond experimentation.

Research Intern at Indian Institute of Technology, Kharagpur (2025-01 – 2025-06)

• Built an end-to-end AI pipeline that predicts a user's personality from multimodal biometric inputs (facial, thermal, radar) and generates a personalized natural-language description.
• Achieved 0.99 SBERT semantic similarity with complete trait coverage across test cases; outperforming multiple frontier LLMs in structural consistency and trait completeness.
• Integrated multilingual voice narration via TTS and translation modules, delivering personality insights in English and Kannada with a ∼30s end-to-end response time on consumer-grade hardware.

AI Intern at Aavaaz AI (2024-12 – 2025-01)

• Built and fine-tuned ASR and NLP models for low-resource Indian languages, enabling real-time voice communication between patients and healthcare providers.
• Improved cross-accent robustness of multilingual speech pipelines through targeted data augmentation and evaluation across diverse speaker demographics, directly reducing transcription error in production.

B.Tech. (Hons.) in Computer Science & Engineering (Data Science) – Chhattisgarh Swami Vivekanand Technical University (2022-11 – 2026-05)