ML Engineer Intern, Dedrone By Axon, Feb 2025 - May 2025

• Building open set recognition methods for wide band spectrograms. Additionally, working on open set object detection.

ML Engineer Intern, Dedrone By Axon, June 2024 - Aug 2024

• Developed track recognition from RF & radar data leveraging Transformer Decoder (Sparse & Linear attention), State Space Models, xLSTM.

• Achieved a MCC of 0.769 with SSM surpassing transformers at 0.71, xLSTM at 0.69 with Focal loss. Reduced model inference time by 90%.

Research Assistant, UMASS IESL & BioNLP lab, Aug 2024 - May 2025

• Enhancing text generation by Learning to Search in Continuous Space. In addition, working on Masked Diffusion language models (MDLM).

• Improving USMLE question generation & answering by integrating self-rewarding, critique, and correction feedback on LLAMA 3.1.

Graduate Student Researcher, Google, Feb 2024 – May 2024

• Improved multi-sample inference on Chain-of-Thought self-consistency and MBR decoding strategies employing arithmetic sampling.

• Attained a 3-5% increase in accuracy on GSM8K (LLaMa-2, Gemma) and 0.45-0.89% increment in COMET score on WMT19 (Flan-T5, MT0).

Research Scientist, Bosch, Aug 2021 – July 2023

• Spearheaded research in responsible AI, focusing on comprehensive vulnerability assessment, robustness, explainability, fairness, and drift detection across diverse domains, including computer vision, time series, speech and language models.

• Pioneered groundbreaking research in AI privacy and security, developing novel attack and defense strategies against various threat models, encompassing adversarial attacks, poisoning, model extraction, and inference attacks.

• Played a pivotal role in the early phases of securing large language models (LLMs), specializing in countering jailbreaking and prompt injection attacks, laying the foundation for the creation of the AIShield Guardian application

• Cultivated strategic partnerships with industry leaders in healthcare, automotive, and financial services, including Whylabs and ClearML, fostering collaborative innovation.

• Translated cutting‑edge research into tangible outcomes, contributing to the filing of four patents and authoring two impactful research papers related to attacks on AI systems

• Architected microservices, CI/CD pipelines, and logging systems for Bosch AIShield product, showcasing expertise in designing robust and scalable systems

Software & DevOps Intern, Fusion Tech, May 2021 – July 2021

• Worked on Server Security and load balancing using Nginx and CSF.

• In addtion, worked on Elastic search in Node.js to build an efficient search engine.

Software development intern, GE Digital, May 2020 – July 2020

• Enhanced GE’s web translation application by implementing cutting‑edge neural sequence‑to‑sequence models including encoder‑decoder with attention, transformers, and BERT.

• Migrated existing translation pipeline based on rules and dictionaries (XML and JSON) to a modern deep learning architecture, boosting translation performance.

• Built and deployed scalable REST APIs with Flask and integrated seamlessly with frontend web interfaces to enable low latency translation.

CMO, TraQez

• Served as Configuration Management Officer (CMO) for Startup Traqez