Spearheaded high-impact client projects across diverse domains, including:1. Developing NLP-powered data pipelines to extract unreported accident-related news articles.2. Building dynamic vehicle control systems for real-time music generation.3. Implementing load forecasting models for efficient energy grid management.4. Optimizing search functionality for e-commerce platforms.

• Developed an end-to-end pipeline with 93% accuracy for detecting access control inconsistencies in android framework images, by integrating code embeddings from Microsoft CodeBert with traditional Machine Learning models.• Implemented a Static Analysis Tool using IBM WALA to identify source code unrelated to API functionality in android system services, resulting in an average decrease of 40% source code across 3000+ methods in android images.• Developed Proof of Concept (PoC) android… Show more • Developed an end-to-end pipeline with 93% accuracy for detecting access control inconsistencies in android framework images, by integrating code embeddings from Microsoft CodeBert with traditional Machine Learning models.• Implemented a Static Analysis Tool using IBM WALA to identify source code unrelated to API functionality in android system services, resulting in an average decrease of 40% source code across 3000+ methods in android images.• Developed Proof of Concept (PoC) android applications to perform comprehensive penetration tests in 50+ system service APIs to evaluate the identified vulnerabilities.• Reported 3 critical vulnerabilities in vendor customised android images for Amazon and LG on the HackerOne platform; more tests are being conducted to identify additional vulnerabilities. Show less

• Assessed over 270 student assignments and tests per term, delivering comprehensive feedback.• Prepared and presented course material to students in small groups of 30 members and in one-on-one sessions.• Dedicated 2 regular office hours per week to support students, addressing their course queries and clarifying course concepts.• Received an “Excellent” rating in my performance evaluation for my commitment and effectiveness in student support.

• Authored 6 research papers which got published in core A and A* rated software engineering conferences.• IJHCI 2022: Developed a survival game in Unity “SurviveCovid-19++” to promote awareness of Covid-19 safety measures, achieving 69.3 quality factor on MEEGA+ evaluation model (excellent).• CSCW 2021: Developed SurviveCovid-19++ : A multiplayer collaborative healthcare game built in Unity3D for educating people about the importance of various roles in preventing the spread of… Show more • Authored 6 research papers which got published in core A and A* rated software engineering conferences.• IJHCI 2022: Developed a survival game in Unity “SurviveCovid-19++” to promote awareness of Covid-19 safety measures, achieving 69.3 quality factor on MEEGA+ evaluation model (excellent).• CSCW 2021: Developed SurviveCovid-19++ : A multiplayer collaborative healthcare game built in Unity3D for educating people about the importance of various roles in preventing the spread of Covid-19• ESEC 2021: Developed the AC2 tool in flask to visualise architectural changes in python projects across release versions.• MSR 2021: Curated a dataset of 526 open-source game engines facilitating empirical studies on game engines.• ASE 2021: Extended the path-based "code2vec" architecture using CFGs and PDGs to improve F1 score by 11% on the full dataset of 16 projects and up to 100% in some individual projects.• CSCW 2020: Developed the "Mood of India" web portal using python, flask, SQL, and C++, performing sentiment analysis on live Twitter posts to visualise the overall sentiment of India in relation to real-time Covid-19 updates. Show less

• Collaborated with 3 distinct research teams to plan and execute end-to-end experiments, while providing technical support for 4+ research projects published in reputed conferences and journals.• Streamlined research workflows using Github CI/CD to automate repetitive technical tasks, reducing project timelines and improving overall project delivery.

• Identified inconsistencies in the performance of the models through the extended evaluation, highlighting the limitations of a restricted evaluation approach.• Analysed the performance of 4 source code retrieval models including BM25 and FastText, by integrating previously unexplored datasets and evaluation methodologies for said models.• Discovered and quantified a 23% increase in retrieval accuracy for a statistical model BM25 compared to neural network model FastText when using MRR… Show more • Identified inconsistencies in the performance of the models through the extended evaluation, highlighting the limitations of a restricted evaluation approach.• Analysed the performance of 4 source code retrieval models including BM25 and FastText, by integrating previously unexplored datasets and evaluation methodologies for said models.• Discovered and quantified a 23% increase in retrieval accuracy for a statistical model BM25 compared to neural network model FastText when using MRR (Mean Reciprocal Rank) evaluation metric.• Demonstrated the significant impact of utilising CodeSearchNet dataset to evaluate FastText algorithm, resulting in a drastic drop from 68.4% to 2.3% for the metric of ‘correct document retrieved at first place’. Show less