Jan 2025 – Jun 2025 · 5 mos

• Designed and implemented an agentic AI pipeline with real-time triggers to autonomously search, deduplicate, and store attack reports,linking together updates about the same attack over time and assessing attack severity, while engineering data models for marketing intelligence and targeted lead generation.This enriched pipeline data feeds directly into a dynamic attack map.
• Architected and developed a voice-interactive agentic AI simulation platform leveraging LiveKit, Django, and OpenAI APIs to deliver ultra-low latency, real-time conversations that interpret attack map data and explain the business impact of disruptions to stakeholders.
• Researched and documented a dynamic Attack Controls Framework ("Guard Framework") using LLMs and knowledge graphs, which automatically populates controls based on previously captured attack data and ensures alignment with industry best practice standards (e.g., NIST, ISO 27001), allowing for flexible addition, deletion, and updating of controls for emerging attack patterns

Jun 2024 – Dec 2024 · 6 mos · London Area, United Kingdom · Remote

• Conducted factor-based financial analysis using Monte Carlo simulations with parameters such as momentum, volume, regime
• filters, and neutralization, addressing survivorship bias and incorporating 10 bps transaction costs to achieve 2-3x returns over the
• S&P 500; leveraged datasets from EODHD, Nasdaq Data Link, and FMP ,resolved data irregularities, analyzed parameter
• relationships,extended to 12 ETFs (XLK, XLU).
• Designed a Gymnasium environment to rank stocks periodically based on historical adjusted close prices, implementing a custom
• convolutional policy with the Stable-Baselines3 PPO algorithm; working on mitigating overfitting issues.

Jun 2024 – Jul 2024 · 1 mo · Chennai, Tamil Nadu, India · Hybrid

• •Developed a hybrid natural selection genetic algorithm and simulated annealing algorithms for optimisation of drone navigation in a multi constrained real world environment reducing travel costs and time, achieving reduction in solution-finding time by 70%

Sep 2023 – Jun 2024 · 9 mos · Mountain View, California, United States · Remote

• Ideated a solution leveraging LLMs to identify phishing threats in email using LLama2 and neural nets that could automatically check the links if present in the email & Analyze content to classify with 87% accuracy, contributed to open-source h2o’s repositories.
• Directed a DLP solution using fine-tuned LLMs to identify and protect PII and financially confidential data, ensuring compliance by NER using the Bert Model; achieved an Impressive Accuracy of 92%

Jun 2023 – Aug 2023 · 2 mos · Ambattur, Tamil Nadu, India · On-site

• Contributed to the 'Life Meter' project to estimate the remaining lifespan of consumables in real-time automated cutting machines, collaborating across departments and delivering a research report to global R&D
• Handled a Time Series dataset of 200GB Utilizing keras and torch to develop Temporal Convolution (TCN) achieving an accuracy rate of 89%
• Achieved this Strategic Feature of Predicting the lifetime that aligns with a real-time minimal resource environment in a cutting machine with an STM32 Microcontroller with 512KB RAM and 2MB Flash

Sep 2022 – Feb 2023 · 5 mos · Hyderabad, Telangana, India · Remote