Adhyatma Chrysostomos Davu
AI & Full-Stack Systems Developer
About Me
I am an Informatics undergraduate at Telkom University (GPA: 3.90/4.00) focused on building AI applications, robust backend architectures, and practical automation systems. My work is anchored in real-world operations: I engineered AuditForge to cut security audit report drafting from 2–3 days to 1–2 hours, shipped 280+ production releases for a live B2B travel and finance platform at PT Eizem Indonesia, and deployed an offline POS with P2P sync running daily on 2 cashier tablets.
I build primarily with Python, Go, PHP (Laravel), and JavaScript, integrating Google Gemini and Groq APIs for production AI features while utilizing Claude Code to accelerate engineering workflows. I am currently seeking software engineering internship or entry-level roles, as well as freelance opportunities to build reliable, high-impact products.
🏆 Achievement & Certifications
- 5th Place out of 50+ teams — Hackattack 2025 National Hackathon (CCI x HIMA IF Telkom University)
-
Microsoft Certified: Azure AI Fundamentals (Score: 964/1000, March 2026)
View Certificate → (Verification Code: wYN4F-2FMo)
- IBM Cognitive Class (October 2026):
AI & Machine Learning
Programming
Web & Backend
Tools
Languages
Professional Experience
Full-Stack Developer (Contract)
PT Eizem Indonesia | Remote, Indonesia- Redesigned NEXORA (eizem.id), a live B2B travel and finance platform used by 15 corporate clients, rebuilding its Laravel 10 interface and booking flows for 6 products, from flights to visas.
- Shipped 280+ production releases with the system owner through staged deploys with backup and rollback, testing changes with Claude Code and headless Chrome before going live.
- Built clear financial records for 15 corporate clients, including transaction history, invoices, e-statements, refund proofs, and printable A4 documents with MDR, FDS, and PPN breakdowns.
- Fixed missing user, division, and department data in 100+ legacy bookings through verified migrations, and set up role-based menus for 4 user types.
AI Analyst Intern
PT Suryasoft Konsultama | Surabaya, Indonesia- Cut audit report drafting from 2–3 days to 1–2 hours by building AuditForge, an LLM pipeline that turns output from 4 security scanners (Nuclei, OWASP ZAP, Burp Suite, Nmap) into reports mapped to CWE, OWASP Top 10, and CVSS v3.1.
- Built a RAG assistant that lets the team ask questions across 200+ internal documents, using hybrid search (vector + BM25) to return grounded, context-aware answers.
- Kept client data safe by masking sensitive information before every LLM call, and made sure auditors approved 100% of AI drafts through a role-based review flow.
- Improved AI-assisted coding and documentation for ongoing company projects by testing LLM outputs and designing prompt pipelines with the AI Analyst team.
Practicum Assistant, Computer Network Course
Telkom University (Informatics Laboratory) | Bandung, Indonesia- Mentored 90+ Informatics students across 3+ Computer Network practicum classes, guiding hands-on configuration and troubleshooting with lecturers and lab teams.
Data & Automation Intern
PT Karya Digital Solusindo | Surabaya, Indonesia- Saved 1–2 hours of manual work per day by automating data workflows in Python, and added new data fields to the company backend for cleaner reporting.
- Analyzed 10,000+ rows of operational data with advanced Excel (VLOOKUP, Pivot Tables, data validation), keeping reports accurate and ready to use.
Education
Bachelor of Informatics
Telkom University | Bandung, IndonesiaCumulative GPA: 3.90 / 4.00
Relevant Coursework: Data Structures & Algorithms, Machine Learning, Web Development, Database Systems
Activities
Selected Core Projects
AuditForge
Addressed manual triage bottlenecks in cybersecurity auditing by building AuditForge, an LLM pipeline that turns output from 4 security scanners (Nuclei, OWASP ZAP, Burp Suite, Nmap) into structured findings mapped to CWE, OWASP Top 10, and CVSS v3.1. Enforces deterministic data masking before LLM calls and cut report drafting from 2–3 days to 1–2 hours with 100% auditor approval through a role-based review workflow.
Personal RAG Assistant
Solved document retrieval hallucinations by engineering a RAG assistant combining ChromaDB vector search and BM25 keyword matching with prompt-injection guardrails. Enabled reliable, context-grounded Q&A with verifiable chunk citations across 200+ internal documents.
Aminah / AMH Techno
Modernized a legacy PHP 5.6 enterprise system by integrating a role-aware AI assistant (Admin, Member, Seller) with document chunking and multi-provider failover (Gemini → Groq). Rebuilt the UI with a dark theme and shipped a companion Progressive Web App (PWA) menu for mobile member access.
Kasir Pregbun
Solved unreliable internet connectivity at a church canteen by building an offline-first POS using UDP broadcast discovery and peer-to-peer Wi-Fi synchronization. Deployed and used daily on 2 cashier tablets to process sales and track inventory conflict-free without internet access.
View Repository →NEXORA (eizem.id)
Redesigned NEXORA (eizem.id), a live B2B travel and finance platform used by 15 corporate clients, rebuilding its Laravel 10 interface and booking flows for 6 products (flights, trains, high-speed rail, ferries, hotels, visas). Shipped 280+ production releases with staged deploys and rollback, generated compliant A4 financial documents (MDR, FDS, PPN), and restored missing records in 100+ legacy bookings.
Log Threat Detection
Addressed web server intrusion blindspots by developing an unsupervised machine learning pipeline using Isolation Forest. Detected anomalous traffic patterns and profiled suspicious actors through an interactive Streamlit monitoring dashboard.
View Repository →TechnoDNA Services
Built the responsive consulting firm website for TechnoDNA Services using Next.js and Tailwind CSS. Showcased technology consulting solutions across web, desktop, IoT, and mobile platforms with optimal load speeds and structured service presentation.
Explore Site →Brain Box
Built a full-stack technical question-and-answer platform similar to Stack Overflow using Ruby. Implemented threaded discourse, voting algorithms, and category tagging for peer problem-solving.
Visit Website →Hospital System Simulation
Simulated patient admissions, doctor schedules, and room management in a healthcare facility using Singly Linked Lists in C++. Modeled queuing bottlenecks and demonstrated deterministic pointer memory management.
View Repository →
































