Projects_

Pelita Technologies submitted this demo for the AmBank HackLab 2.0 Hackathon under Problem Statement 1: Qualitative Market Intelligence Surveillance Solution. Our team is honored to have been awarded 2nd Runner-Up and offered opportunities to further collaborate with AmBank, recognizing the potential impact of our solution. The platform was developed to demonstrate how AI-driven automation can strengthen market surveillance and support better financial decision-making.

The system automates the collection, processing, and analysis of financial announcements, news, and corporate documents to generate evidence-based insights. It leverages web scraping, a structured ETL pipeline, and vector search to transform unstructured data into searchable intelligence. A multi-agent AI layer retrieves relevant information, performs financial analysis, detects risk signals, and returns responses with traceable citations.

Delivered through a FastAPI backend and a Next.js dashboard, the platform provides real-time alerts, an AI copilot, and a comprehensive company view for analysts. Its transparent data acquisition and auditable workflows enhance trust while reducing manual monitoring efforts. Overall, the solution enables continuous, scalable market intelligence to help institutions respond faster to emerging risks and opportunities.

Vouchie is an accountability protocol live on Base and built for Farcaster. Users lock USDC to commit to goals and rely on trusted friends to verify completion instead of centralized moderators. Smart contracts automatically resolve outcomes and distribute funds with no admin control. The system exploits loss aversion, social pressure, and irreversible commitments to drive follow-through. Accountability becomes programmable, social, and enforceable on-chain.

Give Protocol is a DeFi-powered crowdfunding platform that reimagines charitable giving through a no-loss model.

Users stake assets into yield-generating pools using protocols such as Aave and Pendle and allocate a portion of the yield to causes they care about. This allows donors to support meaningful initiatives while their principal continues to grow, creating a sustainable and wealth-building approach to philanthropy.

The project has received multiple recognitions, including 4th place at the BGA Masverse Hackathon and 2nd place in the Morph Buildathon’s Consumer Track. It was also selected for the Scroll Open Campus Season 2 and BGA Incubation Program, where it received a $3k grant at the BGAwards.

This project is submitted for the Alibaba Hackathon. It is an AI-driven platform designed for personalized Malaysian tax relief optimization.

The system uses Computer Vision and OCR on Alibaba Cloud, enhanced with LoRA fine-tuning, to perform intelligent receipt parsing. Qwen handles NLP tasks for producing structured outputs from unstructured receipts.

CAG-RAG, powered by Alibaba Cloud Vector DB and PAI, provides GenAI-driven reasoning to automatically identify eligible tax reliefs. Qwen also supports transaction categorization and powers a conversational tax assistant for user guidance.

A proactive alert engine leverages reinforcement learning to deliver personalized financial advice. The entire platform is built on secure Alibaba Cloud infrastructure, including Function Compute, ApsaraDB, and OSS.

Key AI differentiators include improved OCR accuracy, RL-based auto-tagging, CAG-RAG for deduction discovery, and a GenAI-powered chatbot for intuitive financial interactions.

A decentralized civic-engagement platform that brings Web3 incentives to urban issue reporting. Citizens can submit on-chain petitions for local problems, such as infrastructure damage, while community members stake their support through votes, creating a transparent, market-driven prioritization system.

Petitions are created and stored on the Solana blockchain, providing tamper-proof records that strengthen trust and enable community-driven governance.

This project was built for the Make It Challenge Hackathon and marks my first steps into Web3 development. It was honoured as a top-10 project and awarded a $100 grant.