Summary
Highly skilled AI Software Engineer with comprehensive experience in developing advanced language and machine learning models. Specializes in designing frameworks and solutions for PDF manipulation, legal automation, and model-based detection of offensive content. Passionate about pushing the boundaries of AI applications, from coding support to next-generation PDF frameworks.
I recognize that many roles require experience in fine-tuning existing base LLMs to meet commercial-grade needs. While my direct fine-tuning experience is limited, I have designed and deployed my own advanced model endpoints, focusing on entire model architectures. This approach equips me to effectively fine-tune any model as needed.
Portfolio
Core Skills
- Proficiency in AI Model Development (LLM, BERT-based models)
- Expertise in PDF manipulation frameworks (C++ TFLite)
- Experience developing coding assistants for advanced exploit generation and specialized Googlebot usage
- Web development (Next.js, React)
- Strong background in AGI and multi-model integrations
Experience & Projects
- SageEmail: Automates labeling, organization, and backup of emails (with planned support for additional data types). Classifies emails into routine alerts, critical alerts, important non-alerts, and spam/marketing. Backs up high-priority emails automatically and can provide push or SMS notifications for critical alerts. Offers a web-based UI for custom label definition, advanced AI automation, and user-controlled fine-tuning, initially supporting Gmail.
- Code LLM: Developed and trained an LLM specialized for coding tasks to improve developer productivity.
- Logic LLM: Developed a compact, high-accuracy model optimized for logical problem-solving, enabling efficient local or edge use.
- Safety LLM: Created a general knowledge LLM for open-source learning and experimentation with safety tokens and reinforcement training.
- Exploit Googlebot: Built a gym-based PriorityBFS Googlebot for advanced exploitation proofs-of-concept and penetration testing.
- Exploit LLM: Developed an LLM for generating exploit code and offering in-depth security research capabilities.
- StupidityModel (HateCrimeModel): Implemented a BERT base uncased model to detect hateful content with high accuracy.
- PersonalModel: Built a Bert-base-uncased model as a Chrome addon for personalized AI interactions.
- SageFlower: Engineered a 1024x1024 flower-generation model demonstrating optimized image synthesis under GPU constraints.
- SagePDFFramework: Contributed to a C++ TFLite-powered PDF manipulation framework with performance and feature enhancements.
- SageLegal: Helped develop a multi-model AGI program for automated legal form handling and advanced document analysis.
- TwitchModel: Devised a bert-base-uncased model for categorizing negative or toxic Twitch comments and identifying spam.
- TwitchSpamModel: Specialized sub-model focusing on spam detection and optional comment hiding.
Education
- CS50 (non-completer), Harvard
- Bachelor of Science in Engineering Science, Tufts University
Experience
- AI-Enhanced Software Engineer, PDFSage Inc. (Jan 2025 - Present)
- Entrepreneur (2021 - 2024)
- Product Manager, Kensho Technologies (2013 - 2014)
References
Available upon request.
Work Gaps
I acknowledge gaps in my professional background; my portfolio demonstrates my dedication and capabilities. I value honesty and prefer transparent communication to let the quality of my work speak for itself.