RE:GENESIS — Second-Life Passports for Enterprise E-Waste Recovery on AWS
Built for Environmental Hacks (Waste and Energy / e-waste): how we shipped RE:GENESIS on AWS — detect chassis parts, audit BOM completeness, plan reuse-first disassembly, and mint verifiable Second-Life Passports — plus what fought back on SageMaker cost and vision fallback.
Adaptive electronics recovery — see it, plan it, prove it, reuse it.
Computer vision → BOM completeness → RVS-ranked disassembly → diagnostics → KMS-signed Second-Life Passports
| |
|---|---|
Ship It | Deployed on AWS |
GitHub | github.com/mitra9917/ReGenesis |
Live demo | S3 website UI |
Team | RE:GENESIS — Shubham Mitra & Tanmay |
1. Why this exists — Idea and Impact
Global e-waste reached ~62 million tonnes in 2022. Most still-valuable GPUs, RAM, SSDs, and NICs are shredded for bulk metal recovery — destroying reuse value and locking embodied carbon into landfill economics.
That is a real environmental problem under Waste and Energy: what we throw away, and whether cities and facilities can close the loop instead of burning value.
RE:GENESIS treats a retired server or laptop as a recoverable product, not scrap.
Industrial shredding | RE:GENESIS |
|---|---|
Destroy first, sort metal later | Detect components before they are lost |
No lineage for secondary buyers | Second-Life Passport (KMS-signed, verifiable) |
Opaque ESG claims | Honest LCA ranges with cited factors |
Manual guesswork on teardown order | Reuse Value Score (RVS) plans the sequence |
Who it is for: ITAD operators, circular-hardware recyclers, and reuse marketplaces — the people who decide shred vs recover every day.
Design and usability: someone who is not on our team should open the UI and know the next click — intake → audit → plan → verify.
2. What it does — Sense · Plan · Prove
One visual system. Three modes. No mini-products competing for attention.
Mode | What the operator sees | Outcome |
|---|---|---|
Sense | Intake photo + model key; YOLO + OCR; completeness gaps | Parts found before teardown |
Plan | Reuse Value Score orders disassembly | High-embodied parts first |
Prove | Diagnostics + KMS Second-Life Passports | Buyers verify in one click |
What it does in 60 seconds
- Upload an interior photo — or pick Dell PowerEdge R740 / ThinkPad T14 / Cisco Catalyst 9300
- Vision (SageMaker YOLOv8) or catalog-assisted fallback finds parts and scores BOM completeness
- Step Functions runs detect → parse → plan → diagnose → passport → impact
- Verify a passport:
valid: truevia AWS KMS — tamper the JSON → verify fails
- Impact shows kg diverted + CO₂e avoided as ranges (not fake precision)
3. Working product — vision boxes and RVS plan
On a live recovery run, the operator does not get a black-box score. They get evidence.
What this screen proves
Signal | In the demo | Why judges should care |
|---|---|---|
Detections | 19 components (CPU, RAM, SSD, GPU, PSU, NIC) | Vision sees the chassis before shred |
BOM completeness | Detected vs expected catalog slots | Catches under-detection early |
RVS total | 451.83 — GPU first after cover | Reuse value drives teardown order |
Prerequisites | Step 1 = remove top cover | Safe, operator-realistic sequencing |
Reuse Value Score (RVS) is not FIFO. High-yield components (GPUs, PSUs) surface early; attachment dependencies still win.
BOM means Bill of Materials — the expected parts list for that model. Vision finds what is visible; BOM is the checklist we audit against.
4. Architecture — Built on AWS · Ship It
UPLOAD ORDER: use
06-system-architecture.jpg here.Layer | AWS building blocks | Job in RE:GENESIS |
|---|---|---|
Intake | S3 static UI · API Gateway | Operator front door |
Orchestration | Lambda · Step Functions RecoveryFlow | Auditable spine |
Sense | SageMaker YOLO · Tesseract OCR · BOM catalog | See the parts |
Decide | Completeness audit · RVS planner · diagnostics | Decide what to free first |
Prove | Passport Lambda · KMS · CloudWatch impact | Sign trust + measure diversion |
Persist | S3 · DynamoDB | Images, passports, events |
AWS open source: AWS SAM CLI deploys this stack as infrastructure as code.
Where AWS fits: not a badge — it is the recovery spine.
RecoveryFlow state machine
Hybrid detection is deliberate:
Path | When | Label |
|---|---|---|
SageMaker YOLO | Endpoint up, confidence acceptable | vision |
Catalog-assisted | Low confidence, BOM mismatch, or endpoint OFF | catalog-assisted |
Mock | Local / SAM tests | mock |
How the pieces talk
5. Prove — Second-Life Passports on KMS
Each qualified part gets a signed credential — health grade, lineage, and one-click AWS KMS verification.
Field on the card | Meaning |
|---|---|
KMS SIGNED | Asymmetric signature over canonical JSON |
Health score | Diagnostic qualification (e.g. 80 / 92 / 84) |
Algorithm | ECDSA_SHA_256 |
Verify → | Human action — the system does not auto-approve resale |
Edit the payload after signing → verification returns invalid. That is the trust layer secondary recycling markets need — not a PDF anyone can edit.
6. Execution proof — Pipeline Audit Log
The same trail judges see in the Step Functions console, mirrored in the product UI from DynamoDB PipelineEvents.
What this run earned (live numbers)
Stage | Result | Notable detail |
|---|---|---|
Ingest | Succeeded | Image landed in S3 |
Detect | Succeeded | SOURCE: vision · 19 detections |
Plan | Succeeded | 7 steps · RVS 451.83 · top = GPU |
Test | Succeeded | 19 components tested |
Passport | Succeeded | 17 Second-Life Passports minted |
Impact | Succeeded | 6.24 kg diverted · range 4.55–8.2 kg |
That is execution: a working end-to-end recovery on AWS — not a mock slide.
7. AWS services — named and purposeful
Service | Role | Why this service |
|---|---|---|
AWS SAM / CloudFormation | Whole stack as code | Repeatable facility deploy |
Amazon API Gateway | REST for the UI | One secure front door |
AWS Lambda | Ingest → impact, verify | Pay-per-recovery-job |
AWS Step Functions | RecoveryFlow orchestration | Visual proof; retries; auditable stages |
Amazon S3 | Images, logs, passports, static UI | Durable artifacts + demo URL |
Amazon DynamoDB | Devices, Components, Passports, Events | Low-latency UI polling |
Amazon SageMaker | Custom YOLO endpoint | Domain vision for chassis interiors |
AWS KMS | Sign / Verify passports | Cryptographic reuse trust |
Amazon CloudWatch | Logs + impact metrics | Ops + demo KPIs |
Amazon SNS | Job-complete hook | Operator notifications |
8. What fought back — Learning and growth
- SageMaker billing and wait — An always-on endpoint burns credits. We added explicit on/off/status scripts and refuse to make the environmental workflow depend on a perfect live endpoint.
- Vision fails in the wild — So we shipped catalog-assisted fallback: the pipeline still plans, tests, and signs.
- Trust is a product feature — A PDF “certificate” is not enough. KMS verify plus tamper fail made Second-Life Passports real.
9. Try it
Resource | Link |
|---|---|
GitHub | https://github.com/mitra9917/ReGenesis |
Deployed UI | http://regenesis-dev-uibucket-13iieilvqhoa.s3-website.ap-south-1.amazonaws.com |
Demo video | (paste your YouTube URL) |
Hackathon | Environmental Hacks · Waste and Energy · e-waste |
Closing
Before RE:GENESIS: shred-first, zero reuse credentials.
After: recoverable components, cryptographic lineage, and measurable diversion — so operators can reuse more and throw away less.
RE:GENESIS — recover the component, not just the metal.
— Shubham Mitra & Tanmay · Team RE:GENESIS · Environmental Hacks · Bharat Builds Tour
Image upload cheat sheet (Builder Center)
Upload these files from
docs/builder-blog/assets/ in this order as you scroll the body:Order | File | Section |
|---|---|---|
1 | 01-landing-compare.png | Why this exists |
2 | 02-sense-plan-prove.png | Sense · Plan · Prove |
3 | 03-vision-and-rvs.png | Working product |
4 | 06-system-architecture.jpg | Architecture |
5 | 07-recovery-workflow.png | RecoveryFlow |
6 | 08-sequence-diagram.jpg | How pieces talk |
7 | 04-passports-kms.png | Passports |
8 | 05-pipeline-audit-log.png | Audit log |
Alt text: copy the text inside each
![...]() when the uploader asks for it.Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article