Projects

Explored unsupervised pretraining for task-agnostic representations in multi-task RL; reproduced AVDC results on the impact of shared image-space representations for multi-task learning.

Proposed VideoAgent to self-refine robotic video plans using pretrained VLM feedback and online replanning. Raised MetaWorld success 43.1% → 50%, iTHOR 31.3% → 34.2%, and +22% task acceptance in human evals on BridgeV2.

Built an RL system that uses VLMs to generate reward functions from offline, unlabelled datasets. Achieved a 40%+ gain in success rate and SOTA 0.82 on a real-world assistive-dressing task in out-of-distribution settings.

Built SLAM pipelines (ICP + KD-tree local mapping on CARLA point clouds) for a 20 cm localization gain, and led a 25-student RL & vision team across 3+ international competitions.

Developed visual domain adaptation for CARLA using stable-diffusion models conditioned on simulator images with uniform domain randomization, learning a generalized control policy robust to the sim2real gap.

Led the team to gold with an Active-SLAM + YOLO-World goal-detection pipeline, Hungarian task-agent matching over a sparse bipartite graph, and PRIMAL-2 with hindsight replay for map-agnostic multi-agent path planning.

Built DiffClone for offline, sparse-reward robotic control: a MoCo-fine-tuned ResNet-50 encoder + DDPM behaviour-cloning agent reaching a 92% success rate on pouring, surpassing existing benchmarks.

Closed-domain QA on SQuAD-like data with T5/GPT-3 augmentations and a 3× faster DrQA retriever; an Electra-BERT ensemble hit 0.85 F1 with 2.65× runtime improvement via ONNX + quantization.