Feb 2026 – Present · 4 mos · Hybrid

• Trial-Testing various Diffusion-based Inpainting pipelines (ex. SDXL, ControlNet) from HuggingFace
• models zoo onto entity erasure, appearance change, and entity addition.
• Assembled an open-vocabulary segmentation pipeline using on-premise Google’s OwLViT and Meta’s
• Segment-Anything-Mapping (SAM) to annotate vehicles and pedestrians.
• Experimented on different locally-hosted captioning models (ex. Salesforce’s BLIP, Microsoft’s Kosmos)
• for their understanding on automotive-specific OIs (Objects of Interest).
• So far, got 1 German Patent filed; 1 IEEE IST, and 2 CVIP papers accepted, while others still in progress and reviews!! Continuing to incline myself on R&D works and SOTA Knowledge via scientific literatures and patent documents.

Jun 2024 – Feb 2026 · 1 yr 8 mos · Hybrid

• Proposed an auxiliary fundamental BBox Tracking at the post-processing front (outside of CNN Stack)
• to robustly solve inconsistent object detections by 10%.
• Successfully delivered a proprietary Vision-based Privacy Protection System to prevent other perception
• subsystems from breaching sensitive public identities (accurately upto 95% both in FP32 and INT8
• configurations).
• Enhanced a Sensor degredation-based quality assessment subsystem upto the functional maturity of
• 89% within live tracking of 50 frames per timestamp.
• Continuing to incline myself on R&D works and SOTA Knowledge via scientific literatures and patent documents. So far, got 1 ICASSP'w, 1 ITSC and 2 IEEE IV Symposium papers accepted, while others still in progress and reviews!!

Jun 2023 – May 2024 · 11 mos · Hybrid

• Discussed CNN Architecture improvisations along-with Quantization-aware algorithmic optimizations,
• keeping 3.1ms runtime intact while improving downstream minority class recall upto 6%.
• Analyzed various root causes of critical failures responsible for pipeline’s 10-15% real-life corner scenar-
• ios, from bridging Class Imbalance by 2.6x to mitigating Data distribution-based Bias by 60%.
• Inspected the crucial intricacies of APN’s predictions with their annotations on multiple grounds (Em-
• beddings, MetaData, Stats) via explainable Voxel51’s sandbox tool.
• Developing the key-intuition of technological novelty with the broadened implementations of Computer Vision and AI, via Research-oriented Patent/Paper writings.

Jun 2024 – Aug 2024 · 2 mos · United States · Remote

Jan 2023 – Jan 2023 · 0 mo · Jodhpur, Rajasthan, India · On-site

• Organized the workshop (with Rohan Singh) for 3 days with "Technical Events Team" (head by Nakul Sharma) in "ML-DL series" and conducted the 2nd day about "Deep Learning".

Oct 2022 – Feb 2023 · 4 mos · Jodhpur, Rajasthan, India

• Co-Lead a team of 91 Students into 10 Challenges of diversified domains (ex. ROS, AI, Supply-Chain) by Gov. Agencies (ex. ISRO) and R&D Companies (ex. Consensys).

Aug 2022 – Jun 2023 · 10 mos · Jodhpur, Rajasthan, India

• Worked in "Adversarial Attacks on Deep-Leaning Models" and associating with "Image Analysis and Biometrics Lab" (IAB), head by Prof. Mayank Vatsa and Prof. Richa Singh.

Aug 2022 – Sep 2022 · 1 mo · Jodhpur, Rajasthan, India · On-site

• Helped the respective Boards (BIE and BCCA) to structure and organize the INAE Youth Conclave 2022 for IIT Jodhpur with the core team of around 30 students.

May 2022 – Feb 2023 · 9 mos · Kaiserslautern, Rhineland-Palatinate, Germany

• - Worked on Video-based bi-temporal 2D-3D Change Detection and pivoting Aerial Imagery-based SOTA implementations (Supervised, Unsupervised, or Transfer Learning-based).

Feb 2022 – Mar 2022 · 1 mo · Jodhpur, Rajasthan, India

• Lead a team of 75 Students into 9 Challenges of diversified domains (ex. ROS, AI, Supply-Chain) by Gov. Agencies (ex. DRDO, ISRO) and R&D Companies (ex. BOSCH). Historically breached the Top-10 for the 1st time, with 2 Silver Medals and multiple event-wise Top-10 standings.

Jan 2022 – May 2022 · 4 mos · Jodhpur, Rajasthan, India

• In B.Tech 2nd Year "PRML (Pattern Recognition and Machine Learning)" Course (CSL2050)
• Inculcated the balance of traditional components (ex. Quizzes, Exams) and practical assessments (ex. Kaggle Challenges, Programming Labs) for overall understanding of the course.
• Helped in the batch-wise evaluation of 220 Students, via non-conventional Normalized Marking and Grading Scheme.
• Conducted weekly Lab Sessions and Doubt Addressals, alongside Viva Evaluation to keep a check on fundamental understandings.

Nov 2021 – Apr 2022 · 5 mos · Jodhpur, Rajasthan, India

• Built the designs and worked on mechanisms for Semi-Automatic Shooter(R1) and Picker(R2) Robots to accomplish the definite physical tasks of ”Lagori” (With the team of 45 Members of various domains like ROS, CAD, AI, Electronics).

Oct 2021 – Apr 2022 · 6 mos · IIT Jodhpur

• Brought the culture of ”Research-Papers Group Discussions”, reviewed their SOTA Techniques like Transformers, GANs, etc., and worked on Real-Life Applied Projects (ex. Face-Recognition based Gallery Clustering), focusing on areas such as Object Detection, Modality Translation, etc.

May 2021 – Sep 2021 · 4 mos · Patna, Bihar, India

• Worked on Multi-Modal Multi-Task Learning Framework for detecting COVID-19 Patients, using Chest X-Rays and CT-Scans. Attained every possible evaluation metrics like Sensitivity, Specificity,etc. in the range of 97.5-99% alongside being Generalizable proved via Hypothesis Testing.

Feb 2021 – Mar 2021 · 1 mo · Jodhpur, Rajasthan, India

• Mentored a team of 10 members in WoC 2021 (Winter of Code), working on Open-Source implementions of Machine Learning and Deep Learning Models using only Numpy and Matplotlib.

Aug 2020 – May 2021 · 9 mos · Jodhpur, Rajasthan, India

• Communicated with Professors of various Course-wise Departments (ex. Electronics, Chemical, CS) about Doubts and Feedbacks on UG’23 Electrical Batch, thereby helping to streamline the curriculum throughout the year.

Jun 2020 – Sep 2021 · 1 yr 3 mos · Jodhpur, Rajasthan, India

• Was a Core Team Coordinator in AI-ML Division.