AWS Builder Center
Tracify - Building a Blockchain Investigation & Financial Intelligence Platform on AWS

Tracify - Building a Blockchain Investigation & Financial Intelligence Platform on AWS

Tracify is a blockchain investigation and financial intelligence platform built to simplify the analysis of cryptocurrency transactions and suspicious fund movements. It brings wallet analysis, transaction tracing, fund-flow visualization, evidence management, and AI-assisted insights into a single platform. Powered by AWS services, Docker, APIs, and AI, Tracify helps investigators turn complex blockchain data into structured, actionable intelligence while maintaining secure and traceable evidence handling.

C8HHMB
Tracify - Building a Blockchain Investigation & Financial Intelligence Platform on AWS
PROBLEM
Blockchain investigations are fundamentally broken today. Analysts facing crypto-enabled fraud - from ransomware payouts to romance scams to unregulated exchange abuse - are left with fragmented tools. Block explorers give raw transaction data with no context, spreadsheets fall apart within an hour, and commercial forensic suites are expensive, opaque, and slow to onboard. Meanwhile, victims report incidents through disconnected government portals (NCRP, SAHYOG) that produce data faster than any single investigator can triage. The result: suspicious wallets go untraced, VASP freeze requests take days instead of minutes, and evidence is scattered across screenshots, PDFs, and personal drives. Tracify closes that gap by turning fragmented on-chain data and victim complaints into a single, structured, investigation-ready workflow - with a sealed evidence vault behind it.
WHAT WE BUILT
Tracify is a blockchain investigation and financial intelligence platform for law-enforcement analysts, cybercrime cells, and compliance teams. It ships as a production-grade full-stack application with 104 tests passing, dockerized backend, and cloud deployment on AWS.
Core modules
• Case Management - organize investigations, assignees, priorities and lifecycle status.
• Investigation Workspace - start a trace from a wallet address or complaint; watch a live hop-expansion graph build; inspect any node or edge in a contextual side panel.
• Intelligence Engine -deterministic breadth-first graph expansion, path ranking, signal detection (mixer touchpoints, peeling, structuring, VASP exposure), 0–100 risk scoring with plain-language rationale.
• Complaint Intake - machine-readable ingress (x-api-key) for NCRP / SAHYOG portals; automatic attribution to the nearest highest-confidence VASP; deduplicated alerts with recommended actions.
• Findings & Evidence Vault - record findings with severity and confidence; upload and seal evidence with a SHA-256 chain-of-custody checksum; auto-linked to the parent case .
• Report Generation - one-click LEA-ready attribution reports with CSV export.
• AI Copilot - Gemini-powered natural-language querying of investigation state, summarisation of findings and typology hints.
• Landing Experience - Three.js / react-three-fiber scenes that tell the product story before login.
TECH STACK
Frontend
• React 19, TypeScript, Vite
• TanStack Router & TanStack Query
• Tailwind CSS, shadcn/ui, Radix UI, Lucide icons
• Zustand (client state), React Hook Form + Zod (forms)
• Framer Motion, Three.js, @react-three/fiber (motion & 3D landing)
• Recharts (data visualisation)
Backend
• Node.js 20+, Express 4, TypeScript (strict)
• MongoDB Atlas via Mongoose 8
• JWT + bcrypt authentication, role-based access (investigator / admin)
• Zod request validation, Helmet, rate limiting, NoSQL-injection scrubber
• Structured JSON logger with key redaction and correlation IDs
• Vitest — 104 tests (transport, security, intelligence engine, adapters)
• Docker (multi-stage, non-root user, production dependencies only) Data & intelligence
• Supabase (Postgres) for the frontend metadata layer
• GraphSense adapter for live chain data (BTC / ETH / TRX) with a deterministic synthetic fallback
• Etherscan V2 multichain provider for supplementary EVM data
• Pure, testable intelligence engine (graph expansion, path ranking, signal detection, risk scoring) AI
• Google Gemini API — powers the in-product AI copilot and finding summarisation
WHERE AND HOW AWS IS USED?
AWS is the operational backbone of Tracify's cloud deployment. Four services carry the production workload, backed by least-privilege IAM roles.
  1. Amazon ECS with AWS Fargate - Backend hosting The Node/Express API runs as a Fargate task inside a dedicated Tracify Cluster. The container exposes port 5000 and passes ECS health checks via /api/health. Fargate gives us the container-native deployment model without managing EC2 instances, and lets us scale horizontally when investigation load spikes.
  1. Amazon ECR - Container registry Every backend build is tagged with the Git commit SHA and pushed to the tracify-backend ECR repository. ECS pulls the image via its Task Execution Role - no credentials are baked into the image.
  1. Amazon S3 - Evidence Vault Investigation evidence - PDFs, screenshots, JSON exports, CSVs - is stored as private S3 objects under a strict key convention: • cases/{caseId}/evidence/{uniqueFileName} • cases/{caseId}/reports/{uniqueFileName} • cases/{caseId}/exports/{uniqueFileName} The bucket blocks all public access. Uploads flow through the backend, which authenticates the investigator, authorises against the case, validates file type and size, generates a unique object key, and PUTs to S3. Downloads return short-lived S3 presigned URLs so the browser fetches the file directly without ever seeing AWS credentials. MongoDB stores only the metadata: filename, mime type, size, uploader, S3 key, and SHA-256 checksum.
  1. Amazon CloudWatch - Observability ECS container stdout / stderr streams into the /ecs/tracify-backend log group via the awslogs driver. The structured JSON logger writes request logs, auth failures, S3 errors, and unexpected exceptions - all searchable in CloudWatch Logs Insights. CPU, memory, task health, and container restarts are tracked via CloudWatch Metrics. IAM & security posture Two least-privilege roles: a Task Execution Role for AWS infrastructure operations (ECR pull, CloudWatch write) and a separate Task Role for application permissions (scoped s3:GetObject, s3:PutObject, s3:DeleteObject on the evidence bucket path only). No AWS access keys exist anywhere in code, config, or the Docker image. Development uses the standard AWS credential chain; production uses IAM task-role credentials via the AWS SDK. End-to-end flow User → React frontend → ECS / Fargate API → MongoDB Atlas + Amazon S3, with Amazon CloudWatch observing the backend.
AI CODING TOOLS USED
AI coding tools used Development was accelerated using AI pair-programming and code-generation tools. All AI-generated code was reviewed, tested and integrated by the team; no secrets were ever shared with AI tools. AI handled the pattern-heavy work, while core product decisions - intelligence engine design, evidence chain-of-custody, IAM policy, typology weights - were made by the team.
• Cursor - primary IDE for AI-assisted refactoring, multi-file edits and code review
• Anthropic Claude (Sonnet) - architecture design, backend service-layer implementation, security-hardening review
• Google Gemini - used during development via Cursor and integrated into the product as the AI Copilot backend Credits & license
CREDITS AND LICENSE
Open-Source Libraries
• Runtime: React, Vite, Node.js, Express, TypeScript
• UI: Tailwind CSS, shadcn/ui, Radix UI, Lucide, Framer Motion, Three.js, @react-three/fiber, Recharts
• Data & state: Mongoose, @supabase/supabase-js, TanStack Query, TanStack Router, Zustand, Zod, React Hook Form
• Backend: helmet, express-rate-limit, bcrypt, jsonwebtoken, vitest, tsx Chain-data providers
• GraphSense (Iknaio) — self-hostable open-source blockchain analytics platform
• Etherscan V2 Multichain API AI providers
• Google Gemini API — in-product AI copilot AWS services
• Amazon ECS with AWS Fargate, Amazon ECR, Amazon S3, Amazon CloudWatch (AWS Free Tier where applicable)
LICENSE
Proprietary — all rights reserved. Third-party dependencies retain their original open-source licences (MIT, Apache-2.0, ISC, BSD as declared in package.json).
WHAT WE LEARNED
•Learned how to deploy and manage a backend using AWS ECS, ECR, S3 and CloudWatch.
•Understood the basics of blockchain transaction tracing and wallet analysis.
• Learned how to securely handle investigation evidence using S3 and SHA-256 checksums. Got hands-on experience with Docker, APIs, IAM and cloud security.
• Learned how to integrate AI into a real application using Gemini. Improved our debugging, testing and teamwork while building under a hackathon deadline.
•Most importantly, we learned how to turn an idea into a working, deployable product
Any opinions in this article are those of the individual author and may not reflect the opinions of AWS.
Enjoyed reading this content? Let the author know!

Your likes, comments, shares, and saves help creators reach more builders.

Loading recommendations

Loading article