I build production systems at scale and research how software can become more testable, reliable, and auditable.
I am a Software Engineer at Pathao and an AI researcher based in Dhaka, Bangladesh. My work sits at the intersection of dependable backend systems, graph-based fraud analysis, software engineering, and reliable AI.
- I build Go and Python services that support mobility, food, and parcel products at national scale.
- I work on graph relationship analysis, fraud detection, feature infrastructure, campaign orchestration, and observable distributed systems.
- I hold a BSc in Software Engineering from Shahjalal University of Science and Technology with a 3.92/4.00 CGPA.
- I have solved 2,000+ algorithmic problems across competitive-programming platforms.
I plan to pursue a PhD in Software Engineering, with a particular focus on software testing. I am especially interested in:
- automated test generation and intelligent testing;
- fault detection, debugging, and reliability assessment;
- testing AI-enabled software and LLM-based systems;
- trustworthy and auditable AI pipelines; and
- software engineering methods for reliable code-generation systems.
I am currently working on several projects across these themes and will publish their methods, artifacts, and results as the work matures. I welcome conversations with researchers working on software testing, trustworthy AI, LLM evaluation, and code intelligence.
| Affiliation | Work | Status |
|---|---|---|
| University of Illinois Urbana-Champaign | Co-authoring Discrete Diffusion Models for Code Generation: A Survey of Foundations, Capabilities, Verification, and Evaluation. I lead five of fourteen subsections covering variable-length generation, fixed-canvas limitations, dynamic expansion, hybrid generation, core trade-offs, and open problems. | TMLR manuscript in preparation |
| Concordia University | Co-developed a nine-stage, provenance-preserving LLM pipeline for venture screening and co-designed its validity audit across expert alignment, evidence grounding, outcome alignment, and cross-domain calibration. | Second author; ICAIF 2026 manuscript under review |
- Pathao Loop — campaign orchestration across push, SMS, email, and in-app channels, serving 2.7M+ users daily and saving 3,000+ engineering hours annually.
- Corridor — observable Go reverse proxy centralizing internal ML and data APIs and handling 1M+ calls per day.
- Syndicate & Sherlock — graph-based relationship analysis and multi-domain fraud detection using Go, Neo4j, Django, Celery, PostgreSQL, MySQL, and BigQuery.
- Pathao Itihas — model-ready feature infrastructure connecting historical and real-time data through BigQuery, PostgreSQL, Redis, REST, and GraphQL.
- Prohori — location telemetry and hazardous-route detection using Go, GCP Pub/Sub, Uber H3, and Redis.
- Previously at Truck Lagbe, where I optimized 15+ high-traffic SQL queries by 25–40% and built shared image, messaging, campaign, and authentication services.
| Project | What it does | Stack |
|---|---|---|
| InterviewMate AI | AI-assisted interview preparation with mock interview simulation, video recording, and personalized feedback. | Next.js · MongoDB · OpenAI API |
| EasyCity | Full-stack waste-management platform for DNCC with role-based workflows, tracking, analytics, and billing; a DU Code Samurai national finalist. | React · Express.js · MongoDB |
| Personal Portfolio | Research- and engineering-focused personal website. | HTML · CSS · JavaScript |
| Area | Technologies |
|---|---|
| Languages | Go · Python · C++ · JavaScript |
| Backend | REST · GraphQL · gRPC · Django · FastAPI · Celery · Microservices |
| Data | PostgreSQL · MySQL · Neo4j/Cypher · Redis · BigQuery · Google Cloud Storage |
| Infrastructure | Kubernetes/GKE · Docker · Helm · GitLab CI/CD · Prometheus · Grafana · GCP Pub/Sub · RabbitMQ |
| Research | Software Testing · LLM Evaluation · Diffusion Models · Code Generation · AI Reliability · Graph-Based Anomaly Detection |
- Trustworthy LLM-Assisted Venture Screening: Auditing Expert Alignment, Evidence Grounding, and Cross-Domain Calibration — second author of five; under review at the IEEE International Conference on AI in Finance (ICAIF 2026).
- Discrete Diffusion Models for Code Generation: A Survey of Foundations, Capabilities, Verification, and Evaluation — UIUC Summer Research Program; section lead on five of fourteen subsections; manuscript in preparation for TMLR.
I am open to PhD research conversations and research collaboration, particularly in software testing, dependable software engineering, trustworthy AI, and LLM systems.
- Portfolio: jisan10667.github.io
- LinkedIn: jisan-ahmed-1053651a1
- Email: jisanahmed10667@gmail.com