GAZI SAJID SALMAN@gazi_sajid_salman227
1
Events attended
1
Submissions made
Bangladesh
1 year of experience
About me
I'm a Computer Science student from Bangladesh with a strong interest in backend development and artificial intelligence. I work with PHP, JavaScript, MySQL, and Python, and I'm currently deepening my knowledge of machine learning and deep learning. I've built an IoT predictive maintenance system that combines sensor data with ML models to anticipate equipment failures. I'm passionate about solving real-world problems with data and code, and my goal is to grow into a full-stack and AI engineer. I'm detail-oriented, quick to learn, and enjoy working on projects where technology makes a practical difference. A shorter version, if the field has a character limit: CS student from Bangladesh and backend developer (PHP, JavaScript, MySQL, Python), currently learning machine learning and deep learning. I built an IoT predictive maintenance system using sensors and ML, and I'm working toward becoming a full-stack and AI engineer. If this is for a specific platform, such as an expert or contractor profile, tell me which one and I can tailor it to highlight the skills they care about.
🤝 Top Collaborators
.png&w=640&q=75)
🤓 Latest Submissions
LegacyLift: Safe Java Modernization
PROBLEM Thousands of business-critical applications still run on Java 8 and Spring Boot 2, versions that no longer receive free security updates. Everyone knows they should upgrade, but teams keep postponing it: the original authors are gone, documentation is stale, tests are thin, and one small change can silently break behaviour in production. Migration work is slow, expensive and error-prone. SOLUTION LegacyLift is an agentic modernization workflow powered by IBM Bob in agent mode. It turns an upgrade into four verifiable steps: 1. Explain: Bob reads the whole repository and writes an architecture report: module map, dependency diagram, request flow, migration inventory, risk list and test gaps. 2. Protect: before any change, Bob writes characterization tests that lock in what the code does today. 3. Migrate: Bob upgrades the code on a separate branch under one strict rule: no test assertion may be changed. 4. Prove: all tests run again on the new version. Passing tests mean behaviour was preserved, not that tests were adjusted. RESULTS (Spring PetClinic, a real open-source app) - Java 8 to Java 21, Spring Boot 2.7.3 to 3.4.5 - Tests 41 to 71 (30 characterization tests written by Bob); branch coverage of the critical Owner and PetValidator classes raised to 100% - 14 files migrated from javax to jakarta - All 71 tests run on Java 21 with 0 failures - 1.7 hours end to end, versus an estimated 24 hours of manual work (about 93% less time) We verified independently that test logic is identical before and after migration. Human review also caught mistakes in Bob's first analysis (it missed 4 of 14 files); we fixed the prompt and documented the corrections. Trust the agent, verify with evidence. TARGET USERS AND HOW THEY USE IT Developers and teams maintaining legacy Java/Spring systems. They open their repository in IBM Bob and run the versioned prompts in order (PROMPTS.md for PetClinic, PROMPTS_TEMPLATE.md for any Java/Spring project), reviewing each report before the next step. A Streamlit dashboard shows the results: KPIs, coverage and effort charts, all reports, and a Bob Evidence tab with every Bob session. WHAT MAKES IT UNIQUE Most AI migration tools just rewrite code. LegacyLift makes the agent write the safety net first, forbids it from touching that safety net, and produces evidence (reports, coverage, test diff checks, metrics) that the result behaves like the original. It also keeps a human-in-the-loop checkpoint after every step. COMPANION TOOL We also packaged the mechanical steps into LegacyLift Converter: upload a legacy project zip and download it migrated with a test-verified report (OpenRewrite plus fixes learned in this project). For transparency, the converter was built with Claude Code, not IBM Bob; all analysis, tests, migration, reports and the dashboard were done with IBM Bob.
27 Sep 2026
