Trustworthy AI and RAG
Reducing hallucinations, grounding generated answers, and evaluating retrieval quality.
Research and experimentation
AI and Software Engineering Research
Md Soyebuzaman Naim explores trustworthy AI, retrieval-augmented generation, HealthTech, FinTech, robotics, and production software. The work connects undergraduate research with systems that can be tested, documented, and used beyond a demo.
Research focus
HealthTech and FinTech
Exploring software and AI systems for consequential, real-world problem domains.
Robotics and production systems
Connecting sensing, edge intelligence, backend architecture, and reliable delivery.
Research work and roles
Technical Lead
IEEE Computer Society SEU Student Branch Chapter
โข Lead and support technical workshops, seminars, and student initiatives involving software engineering, AI, cybersecurity, and emerging technologies. โข Coordinate technical activities and contribute to planning student-focused learning, innovation, and community programs.
Successfully organized and delivered hands-on technical initiatives and hackathons including the AI & API Hackathon 2026.
Undergraduate Researcher
Southeast University
โข Exploring applications of Retrieval-Augmented Generation (RAG) and large language models in healthcare and financial technology domains. โข Investigating domain-aware retrieval, information grounding, hallucination reduction, and reliable AI-assisted systems using external knowledge sources.
Awarded Best Emerging Idea at SciBlitz 2.0 (IEEE CUET) for Trust Layer concept in RAG systems.
Awards and evidence
๐ Secured ๐ฐ๐๐ต ๐ฃ๐น๐ฎ๐ฐ๐ฒ in the ๐ฃ๐ฟ๐ผ๐ด๐ฟ๐ฎ๐บ๐บ๐ถ๐ป๐ด ๐๐ผ๐ป๐๐ฒ๐๐ at ๐ฆ๐๐จ ๐๐ฆ ๐๐ฎ๐ฟ๐ป๐ถ๐๐ฎ๐น ๐ฎ๐ฌ๐ฎ๐ฒ!
Sat for a coding contest today after quite a long break. I actually joined a bit late, so I wasn't expecting much, but I managed to finish in the 4th position. Honestly, more than the rank, it just felt great to get back into the problem-solving vibe. There's something really satisfying about racking your brain over a logic issue and finally getting the code to run under time pressure. It was a much-needed confidence boost to get back on track. Looking forward to participating in more of these. Thanks, Southeast Computer Club for organising such a wholesome event.
Champion - AI & API Hackathon 2026 (First Edition)
Won 1st Place with Team "Error 429: Too Many Devs" building Civic AI โ an AI-powered civic issue reporting and government management platform with resilient Gemini/Groq failover, PII isolation, and geospatial deduplication.
Best Emerging Idea Award โ SciBlitz 2.0
Served as Team Lead for Team Zero Volt in the 3-Minute Thesis segment organized by IEEE CUET Student Branch, presenting a novel Trust Layer concept for reducing RAG hallucinations.
Codeforces Competitive Programmer (Handle: Naiminator)
Solved 340+ algorithmic problems on Codeforces with maximum rating of 1123. ICPC Preliminary Contest participant in 2024 and 2025.