Sep 2024 – Dec 2024 · 3 mos · Riverside, California, United States · On-site

• 1) Spearheaded course development for EE 148 - Introduction to Robotics and Artificial Intelligence (1st Edition) at the University of California, Riverside.
• 2) Mentored 20+ undergraduate students on Computer Vision, ROS, and Machine Learning topics.
• 3) Assisted in setting assignments, grading exams, and conducting office hours to support the professor's teaching efforts.

Jun 2024 – Aug 2024 · 2 mos · Wilmington, Delaware, United States · On-site

• 1) Developed a novel reject inference-based ML model for credit card applications, improving credit performance assessment.
• 2) Established MLOps pipeline on Azure DevOps and AWS for model training, deployment, and drift monitoring.
• 3) Incorporated human-in-the-loop evaluations and refined models based on proprietary KPIs. Improved the overall performance by 6% while utilizing 50% less training data.

Oct 2023 – Aug 2025 · 1 yr 10 mos · Riverside, California, United States · On-site

• Conducting graduate-level research on topics related to human pose estimation, vision-language models, and deepfake detection under Dr. Amit Roy Chowdhury. Here are some of my recent projects which led to publication (first/co-first author):
• 1) Human Pose Estimation under Occlusion: Proposed a novel Test-Time Training (TTT) strategy for performing pose estimation under occlusion. Paper published at WACV 2025 Oral.
• 2) Safety Alignment of Vision-Language Models: Discovered a novel image-encoder early-exit-based vulnerability and proposed a layer-wise Proximal Policy Optimization (L-PPO) method based on RLHF to address it effectively. Paper accepted as ICML 2025 Spotlight Poster.
• 3) Novel Occlusion-based Pose Estimation Dataset: Proposed a new video benchmark dataset for 3D human pose and shape estimation under real occlusions using Gaussian Splats/NeRFs and 3D graphics software such as Blender and the Unity game engine. Paper accepted at ICCV 2025.

May 2022 – Aug 2023 · 1 yr 3 mos · Hyderabad, Telangana, India · On-site

• Conducted full-time state-of-the-art research on topics related to efficient transfer learning, causal learning, and multimodal learning in weak supervision settings under the guidance of Dr. Vineeth Balasubramanian. Below are the publications resulting from this work over a span of 16 months:
• 1) Causal Discovery using LLMs: Proposed novel algorithms to perform causal discovery using large language models like Phi-3 and Llama-3. Paper published at ICLR 2025.
• 2) Transferability Estimation of Ensemble Models: Proposed a novel information-theoretic and optimal transport-based metric for estimating fine-tuning performance of model ensembles. Paper accepted at ICCV 2023.
• 3) Inducing Causal Effects for Neural Network Training: Built in direct and indirect causal effects into neural network training to make them interpretable and reliable. Paper published at AAAI 2024.

Nov 2021 – May 2022 · 6 mos · Bengaluru, Karnataka, India · Remote

• 1) Developed two novel lane marking detection algorithms for level 2 ADAS, improving F1-score by 5%.
• 2) Established MLOPs pipelines for deployment, monitoring, and re-training on in-house test beds.
• 3) Published papers at international machine learning conferences like ICPRS 2022 and INISTA 2022.

Sep 2021 – Nov 2021 · 2 mos · Osnabrück, Lower Saxony, Germany · Remote

• 1) Collaborated with the Institute of Cognitive Science to study information processing in multi-path neural networks.
• 2) Co-authored a paper published at the ICLR Workshop 2021, highlighting insights into cognitive-inspired neural architectures.

May 2021 – Dec 2021 · 7 mos · Kaiserslautern, Rhineland-Palatinate, Germany · Remote

• 1) Worked on high-impact projects focused on skin lesion detection using image classification techniques for the early diagnosis of skin cancer, under the guidance of Dr. Andreas Dengel and mentorship of Dr. Adriano Lucieri.
• 2) Developed synthetic skin lesion images using generative models such as StyleGAN, incorporating concept-based variations to enable interpretable image classification for skin lesion detection.

Apr 2020 – Aug 2021 · 1 yr 4 mos · Nagpur, Maharashtra · On-site

• 1) Founded and built a student-led marketplace app enabling students to buy, sell, and rent daily-use items.
• 2) Led full-stack development, design, outreach, and business growth, onboarding a large portion of the student community.
• 3) Won 3 startup competitions worldwide.

Aug 2019 – May 2022 · 2 yrs 9 mos · Nagpur, Maharashtra, India · On-site

• 1) Led and contributed to several projects in computer vision and robotics, while also mentoring junior undergraduates on applied CV and ML research.
• 2) Served as Treasurer of the IEEE VNIT Student Branch, actively supporting tech initiatives and organizing knowledge-sharing sessions.

Jun 2019 – Apr 2020 · 10 mos · Nagpur, Maharashtra, India · On-site

• 1) Designed and implemented a face liveness detection pipeline to distinguish between real and spoofed faces using deep learning techniques.
• 2) Trained and evaluated CNN-based models using PyTorch and Keras, incorporating real-time preprocessing with OpenCV for frame extraction and augmentation.
• 3) Deployed the solution on Azure for scalable inference and integrated model monitoring and performance tracking.