
ScrapeSaathi: Bringing India’s Informal Collectors into the Formal E-Waste Chain with AWS
Transforming informal last-mile e-waste collection into a verifiable, CPCB-compliant circular economy network using Amazon Bedrock, Serverless Lambda, and IoT-driven custody verification.
ScrapeSaathi: Bringing India’s Informal Collectors into the Formal E-Waste Chain with AWS
Author: Navya Minocha
Project Repository: GitHub - Scrape-saathi
Regulatory Framework: CPCB E-Waste Management Rules, 2022
The Core Problem
Over 90% to 95% of India’s 1.7 million tonnes of annual electronic waste is aggregated by informal collectors (kabadiwalas and waste-pickers) who hold unmatched last-mile collection efficiency but remain completely locked out of the formal Extended Producer Responsibility (EPR) recycling chain under the E-Waste (Management) Rules, 2022 .
The Backyard Hazard Crisis
Excluded from formal channels, collectors are forced to rely on hazardous backyard processing, open-air cable burning that releases cancer-causing dioxins and furans into urban air, crude nitric acid and cyanide leaching of circuit boards that destroys local groundwater, and hammer-crushing of CRT tubes and batteries that causes vacuum implosions, lead dispersion, and thermal runaway fires.
Loss of Strategic National Minerals
Primitive backyard extraction only recovers basic copper and gold scrap while burning off and permanently losing India's strategic critical raw materials, including Lithium, Cobalt, Neodymium, Tantalum, Gallium, and Indium, which are vital for domestic green manufacturing.
The Intermediary Exploitation Gap
The root bottleneck is informational and economic rather than operational; informal collectors do not know prevailing factory-gate commodity rates and are exploited by informal aggregators and middlemen who capture a 30% to 45% margin haircut on every collected kilogram.
The Idea of ScrapeSaathi
We designed an ultra-low-bandwidth, vernacular, audio-narrated mobile and edge platform that enables informal collectors to discover fair market prices, catalog digital lots, match directly with nearby CPCB/SPCB-authorized recyclers, execute a legally verified handover, and receive immediate cash or UPI settlements. Source code is open on GitHub .
Radical Differences from Conventional Solutions
- No AI Weight Hallucination: Unlike theoretical systems that claim to predict scrap weight from smartphone photos, our platform completely rejects 2D computer vision weight-guessing because optical images cannot detect internal density, wall thickness, or component voids. Instead, the collector explicitly logs actual measured weight from their physical hanging scale (kanta/tarazu) using large-dial, sunlight-readable, single-thumb quick-add buttons (+1kg, +5kg, +10kg, +25kg).
- Low-Literacy Audio & Vernacular Core: The interface is built from the ground up in Hindi and Marathi alongside English, coupled with native speech synthesis ( Amazon Polly / Web Speech API) that speaks live market prices, lot valuation breakdowns, and safety warnings aloud so that collectors with zero literacy can use the app without assistance.
- Cash-First Frictionless Handover: Understanding that requiring immediate bank transfers, GST registrations, or tax compliance alienates daily-wage collectors, the platform treats immediate cash payout as a first-class citizen, letting authorized recyclers disburse hard cash on the spot while generating a legally binding digital transfer record.
- Zero-Cost Collector Model: The platform is permanently 100% free for informal waste collectors; operational costs are fully sustained by authorized recyclers who pay a 2% digital verification fee per lot to secure audit-ready, CPCB-compliant EPR credits.
How AWS Cloud Powers ScrapeSaathi
- Serverless Scale with AWS Lambda & Amazon API Gateway: To handle sudden bursts of thousands of collectors accessing price boards during morning market hours without paying for idle server infrastructure, we engineered the core backend on AWS Lambda and Amazon API Gateway using
FastAPIandMangum, delivering elastic auto-scaling with sub-50ms execution latency and zero server management overhead.
- Multimodal Scrap Triage Powered by Amazon Bedrock: We deployed Amazon Bedrock leveraging multimodal foundation models (
Amazon Nova LiteandAnthropic Claude 3.5 Sonnet) to inspect scrap photos, classify material grades, and flag severe visual hazard markers (swollen lithium battery pouches or unstripped cables) without any manual pipeline configuration or GPU provisioning.
- Sub-Millisecond Geospatial Recycler Matching via Amazon Aurora PostGIS: We utilized Amazon Aurora Serverless running PostgreSQL with the PostGIS spatial engine to execute indexed spatial distance queries (
ST_DWithinandST_Distance_Sphere), calculating the shortest transit routes between a collector’s GPS coordinates and authorized recycling depots with active CPCB licenses within single-digit milliseconds.
- Instantaneous Commodity Price Discovery on Amazon DynamoDB: Daily scrap price fluctuations across regional industrial hubs (Mumbai, Pune, Nagpur, Delhi-NCR) are stored in Amazon DynamoDB , ensuring that low-end Android smartphones operating on spotty 2G/3G connections can fetch prevailing market rates with sub-5 millisecond read latency.
- Tamper-Evident Photo & Audit Vault Using Amazon S3: Handover photographs, weighbridge tickets, and inspector logs are archived directly in Amazon S3 utilizing Server-Side Encryption (
SSE-KMS) and Write-Once-Read-Many (WORM) Object Lock policies, establishing a non-repudiable legal repository that satisfies environmental regulatory inspections.
- Real-Time Hardware Weighbridge Integration via AWS IoT Core: To prevent weighbridge fraud and scale tampering during material handovers, recycler-side digital scales publish live tare and gross weights over MQTT to AWS IoT Core , streaming certified sensor payloads directly into the platform ledger without manual human intervention.
- Cryptographic CPCB EPR Verification via AWS Key Management Service (AWS KMS): Every verified handover generates an immutable SHA-256 certificate token hashed using AWS KMS , mathematically binding the lot reference code, collector ID, recycler CPCB license number, certified scale weight, total payout, and GPS timestamp into an unforgeable QR verification slip.
Direct Economic & Environmental Impact
- Direct Unit Economic Uplift for Collectors: On a standard 70 kg mixed lot (15 kg PCBs, 25 kg copper cables, 10 kg Li-ion packs, and 20 kg copper motors), an informal middleman pays approximately ₹21,950, whereas ScrapeSaathi’s direct recycler channel secures ₹30,600, putting ₹8,650 of net extra cash (+39.4% income increase) directly into the collector's pocket.
- Elimination of Open Cable Burning: Diverting informal scrap into authorized mechanical granulators and zero-emission hydrometallurgical facilities prevents thousands of kilograms of carcinogenic open PVC cable burning.
- Groundwater & Community Protection: Halts primitive nitric acid and cyanide leaching of circuit boards, protecting municipal water tables from toxic heavy metal contamination while eliminating violent CRT implosions and battery fires.
- Critical National Mineral Sovereignty: Recaptures vital battery metals and rare earth elements back into India's domestic circular manufacturing economy.
Explore the Project
- Live Repository: github.com/ravishkansal22/Scrape-saathi
- Community Profile: Navya Minocha on AWS Community profile
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article