I am a Software Engineer and AI Researcher working at the intersection of Web Security & Privacy, Machine Learning, Static Application Security Testing (SAST), Dynamic Application Security Testing (DAST), and Large Language Models (LLMs). Currently, I build security-focused ML pipeline infrastructure, observability, and threat-modeling frameworks at SS Software Solutions in Hyderabad.
I completed my Master of Technology in Data Science & Artificial Intelligence at the Indian Institute of Information Technology (IIIT), Ranchi (CGPA: 8.79/10), supported by the national IETE Postgraduate Scholarship. My research investigates parameter-efficient fine-tuning (LoRA), catastrophic forgetting mitigation with Elastic Weight Consolidation (EWC), and adversarial prompt injection defenses against cryptographic secret leakage. I was awarded the Best Paper Award at ICACA 2026.
I am actively seeking PhD opportunities to advance research in secure, robust, and reliable foundation model systems.
News
Apr 2026Served as Peer Reviewer for ICICST 2026 (NIT Jalandhar) — Certificate.
Mar 2026Joined SS Software Solutions as Software Engineer focusing on ML infrastructure and governance.
Jan 2026Received Best Paper Award at ICACA 2026 for research on LLM cryptographic key leakage under prompt injection.
2026Paper on AST-Enhanced Code Review Assistants accepted in Springer Nature (LNNS).
Jul 2025IoT-HITS paper published in IETE Journal of Research (Taylor & Francis).
Mar 2022Smart City Accident Monitoring published in IETE Journal of Research (Taylor & Francis).
Publications
Adversarial Prompt Injection Attacks on Large Language Models: Cryptographic Key Leakage and Defense Strategies
Ravi Prakash Srivastava, Sourabh Sah, Kirti Kumari, Pavan Vovveti, Md. Shah Fahad
Proceedings of the International Conference on Advances in Computing and Applications (ICACA 2026), Springer Nature LNNS [In press]
Best Paper Award
Demonstrates multi-turn context hijacking vectors leaking system prompts and cryptographic keys, proposing instruction hierarchy boundaries and input sanitization defenses.
@inproceedings{srivastava2026adversarial,
title={Adversarial Prompt Injection Attacks on Large Language Models: Cryptographic Key Leakage and Defense Strategies},
author={Srivastava, Ravi Prakash and Sah, Sourabh and Kumari, Kirti and Vovveti, Pavan and Fahad, Md. Shah},
booktitle={Proceedings of the International Conference on Advances in Computing and Applications (ICACA 2026)},
publisher={Springer Nature, Lecture Notes in Networks and Systems},
year={2026},
note={Best Paper Award}
}
Deepfake Video Detection Using Face-Centric Processing and Frame Sampling
Sourabh Sah, Ravi Prakash Srivastava, Kirti Kumari, Md. Shah Fahad
Proceedings of the International Conference on Advances in Computing and Applications (ICACA 2026), Springer Nature LNNS [In press]
Introduces targeted facial landmark sampling and temporal frame reduction to reduce inference overhead while preserving detection accuracy on synthetic video streams.
@inproceedings{sah2026deepfake,
title={Deepfake Video Detection Using Face-Centric Processing and Frame Sampling},
author={Sah, Sourabh and Srivastava, Ravi Prakash and Kumari, Kirti and Fahad, Md. Shah},
booktitle={Proceedings of the International Conference on Advances in Computing and Applications (ICACA 2026)},
publisher={Springer Nature, Lecture Notes in Networks and Systems},
year={2026}
}
AI-Powered Code Review Assistant: Eliminating LLM Pipeline Using AST-Enhanced Hallucination to Analyze Pull Requests Automatically
Sairamakrishna BuchiReddy Karri, Ravi Prakash Srivastava, Kiran Kumar Kadimi Chenchu, Ravina Marilyn Banze, Ryan Banze
Proceedings of the International Conference on Network Security and Blockchain Technology (ICNSBT 2026), Springer Nature LNNS [Accepted]
Combines Abstract Syntax Tree (AST) validation with LLM code analysis to catch syntax and logic flaws automatically while suppressing false positive hallucinations.
@inproceedings{karri2026aipowered,
title={AI-Powered Code Review Assistant: Eliminating LLM Pipeline Using AST-Enhanced Hallucination to Analyze Pull Requests Automatically},
author={Karri, Sairamakrishna BuchiReddy and Srivastava, Ravi Prakash and Chenchu, Kiran Kumar Kadimi and Banze, Ravina Marilyn and Banze, Ryan},
booktitle={Proceedings of the International Conference on Network Security and Blockchain Technology (ICNSBT 2026)},
publisher={Springer Nature, Lecture Notes in Networks and Systems},
year={2026}
}
Ravi Prakash Srivastava, Sanjay Dahiya, Aashish Nehra
IETE Journal of Research, Taylor & Francis, Vol. 71, Issue 11, July 2025
Multi-sensor edge perimeter intrusion detection system combining lightweight deep learning classification with resilience to adversarial environmental noise.
@article{srivastava2025iothits,
title={IoT-HITS: An IoT Based Human Intrusion Detection System for Border Region Using Deep Learning},
author={Srivastava, Ravi Prakash and Dahiya, Sanjay and Nehra, Aashish},
journal={IETE Journal of Research},
volume={71},
number={11},
year={2025},
publisher={Taylor \& Francis},
doi={10.1080/03772063.2025.2530182}
}
@article{kumar2022smartcity,
title={Smart City Vehicle Accident Monitoring and Detection System Using (MEMS, GSM, GPS) Raspberry Pi 4},
author={Kumar, Ajay and Akhtar, Mohammad Amir Khusru and Pandey, Abhishek and Srivastava, Ravi Prakash},
journal={IETE Journal of Research},
volume={69},
number={11},
pages={8121--8129},
year={2022},
publisher={Taylor \& Francis},
doi={10.1080/03772063.2022.2043787}
}
High-throughput semantic caching layer routing natural language prompts via vector embeddings rather than rigid string matches, trimming API inference cost and latency by ~40%.
Vulnerability management platform integrating Static (SAST) and Dynamic (DAST) Application Security Testing for Web applications. Automated static code analysis with Semgrep for early detection of security vulnerabilities, and built an interactive Streamlit dashboard for real-time visualization, severity classification, and remediation recommendations.
Standardized end-to-end machine learning infrastructure with MLflow experiment tracking, automated schema validation, containerization, and GitHub Actions CI/CD deployment.
Python · MLflow · Docker · GitHub Actions · Flask
Cross-Domain PEFT Summarizer (Master's Thesis)
Adapts BART-large-CNN from news summarization to legal documents (BillSum) using LoRA combined with Elastic Weight Consolidation (EWC) to preserve pre-trained knowledge, fine-tuning under 1% (~2M parameters) of the model.
Financial anomaly detection pipeline benchmarked across ensemble classifiers handling extreme class imbalance via SMOTE, optimizing cost-sensitive precision-recall trade-offs.
Python · Scikit-learn · SMOTE · Random Forest
Experience
Software EngineerMar 2026 – Present
SS Software Solutions LLC · Hyderabad, India
Built security-focused ML infrastructure: drift detection and observability monitoring for production models, JSON-schema structured logging, and model versioning (MLflow, DVC), making model degradation under changing data visible early and cutting incident diagnosis time by ~40%.
Applied static analysis for vulnerability detection in GitHub Actions CI/CD: SAST (Semgrep, Bandit) and dependency scanning to catch vulnerable code patterns before deployment, reducing security-review effort.
Performed STRIDE threat modeling of ML workflows, identifying attack surfaces (data, model, and pipeline) and embedding security and governance controls into the development lifecycle.
Machine Learning InternAug 2022 – Feb 2023
Centre for Development of Advanced Computing (C-DAC), Govt. of India · Hyderabad, India
Developed supervised ML models to predict student academic performance from educational data.
Implemented and benchmarked Decision Tree, Random Forest and Support Vector Machine classifiers, building the experimental evaluation skills used in my later research.
Education
Master of Technology (M.Tech) in Data Science & AI2024 – 2026
Indian Institute of Information Technology (IIIT), Ranchi, India · CGPA: 8.79 / 10
Recipient of the national IETE Postgraduate Scholarship (2024–2026) for academic excellence.
Thesis: Cross-Domain Text Summarization Using Transformer Models.
Bachelor of Technology (B.Tech) in Information Technology2018 – 2022
North-Eastern Hill University (NEHU), Shillong, India · CGPA: 7.85 / 10
Awarded Associate Membership (AMIETE) of the Institution of Electronics and Telecommunication Engineers (IETE), India (Since 2022).
Thesis: IoT-HITS: Edge Deep Learning Intrusion Detection for Border Perimeters.
