Medha Hira

CV

Download PDF ↓

Education

  • M.S. in Computer Science & Natural Language Processing, Carnegie Mellon University — Pittsburgh, PA (August 2025 – December 2026)
    • Language Technologies Institute, School of Computer Science
    • Research: retrieval for code-switched and low-resource languages (Advisor: Dr. Maarten Sap)
    • Coursework: Advanced Natural Language Processing, Deep Learning Systems, Advanced Introduction to Machine Learning, Search Engines
  • B.Tech in Computer Science and Engineering, Indraprastha Institute of Information Technology (IIIT) Delhi — New Delhi, India (September 2021 – June 2025)
    • CGPA: 9.22/10.00
    • Teaching Assistant: Machine Learning
    • Dean’s List for Academic Excellence: 2023, 2024
    • Coursework: Data Structures, Algorithm Design and Analysis, Operating Systems, Machine Learning, Natural Language Processing, Information Retrieval, Deep Learning, Large Language Models

Work Experience

  • Software Engineer (ML) Intern, Google — San Jose, USA (May 2026 – Present)
    • Developing automated quality hill-climbing pipelines for Gemini Enterprise, including intelligent test-case generation, failure discovery, and autonomous fixer agents that iteratively improve system reliability and response quality.
  • Research Intern, Microsoft Research India — Bangalore, India (January 2025 – June 2025)
    • Supervisor: Dr. Akshay Nambi
    • Enhanced Microsoft Copilot’s Agentic Deep Research reasoning and planning module through a novel chain-of-thought and custom probing framework, boosting user intent understanding and plan quality by 29%.
    • Developed a novel benchmark to assess unsafe tool-use in LLM agents and designed a novel reward function for training agents using RL (GRPO), reducing unsafe agentic actions by 20% while retaining normal use-case efficacy.
  • Software Engineer Intern, Google — Bangalore, India (May 2024 – August 2024)
    • Designed and deployed a scalable custom ranking algorithm for Google TV, tailored to product-specific KPIs. The algorithm improved search relevance and significantly boosted user engagement metrics, and was successfully dogfooded and launched to production, enhancing personalized content discovery for millions of users.
    • Built a modular and extensible ML experimentation framework in Java to accelerate model development cycles by 40%, enabling seamless configuration swaps and streamlined A/B testing across ranking pipelines.
  • Data Science Intern, SAP — Dubai, U.A.E. (July 2023 – August 2023)
    • Engineered a Selenium-based Python bot to automate the ETL pipeline for large-scale employee data, incorporating data cleaning and validation to improve reliability and eliminate manual intervention.

Research Experience

  • Undergraduate Researcher, Infosys Center for Artificial Intelligence, IIIT-Delhi — Delhi, India (September 2023 – April 2024)
    • Supervisor: Dr. Anubha Gupta
    • Developed a prosody-preserving speech-to-speech translation system leveraging Whisper-Fast, Meta’s MMS-TTS, and X-vector voice cloning. Achieved a 15% improvement in MOS and +8 BLEU compared to direct S2ST systems.
    • Proposed a co-attention multi-task learning framework in PyTorch for speaker emotion recognition, outperforming SOTA models with an 8% F1 gain on unseen multilingual speaker profiles.
  • GRM Programme Mentee, IBM Research — Remote (August 2023 – October 2023)
    • Supervisors: Dr. Sameep Mehta, Dr. Nishtha Madaan
    • Built a modular LLM guardrail ensemble for enterprise contexts, supporting topic filtering (87% via LDA and keyword methods), toxicity detection (91–95% with bi-encoders), and hallucination mitigation through RAG-based grounding.

Publications

Honors and Awards

  • Rank 6, Amazon ML Challenge (2024)
  • Summer Undergraduate Research Fellowship, IIIT-Delhi (2023)
  • IBM Global Remote Mentoring Programme (2023)
  • Selected Researcher at IIIT-Hyderabad via the SRISHTI Programme (2023)
  • Women Engineers Program by TalentSprint, supported by Google — top 250 out of 25,000 applicants (2022)
  • Google Developers ML Bootcamp — 72 students selected (2022)

Technical Skills

  • Languages: Python (NumPy, Pandas, Matplotlib, Scikit-learn, SciPy), C, C++, Java, SQL, HTML/CSS, JavaScript, Bash
  • ML/DL: PyTorch, TensorFlow, OpenCV, NLTK
  • GenAI: LangChain, OpenAI, HuggingFace, vLLM, VeRL, Unsloth
  • Databases/RAG: MySQL, Pinecone, MongoDB, LlamaIndex
  • Other: Docker, Git, Linux, Django, REST, Streamlit

Teaching

  • Teaching Assistant, Machine Learning — IIIT-Delhi