Who you work with#
SELCO Foundation is a Bengaluru-registered trust working on sustainable energy access for underserved communities, using decentralised renewable energy to reduce poverty and improve livelihoods. It works across agriculture, animal husbandry, textiles and crafts, micro-businesses, healthcare and the built environment, with documented work in Karnataka, Odisha, Jharkhand and Assam.
This request comes from its Innovation Team, which designs the energy-enabled livelihood work and needs to understand a value chain before it can place a solution inside it.
The situation#
Before SELCO can decide where a solar dryer or a cold room actually changes someone’s income, it needs to know how a value chain works end to end — who does what, where value is created and lost, where energy and technology are missing, and where a farmer or artisan could move up into being an enterprise.
SELCO needs this for 32 value chains, of which 13 are the priority: rice and paddy, turmeric, millet, pineapple, mushroom, honey, IoT irrigation, poultry, piggery, cane and bamboo, silk, eateries and cafes, and snack making. Each one becomes a presentation following a fixed ten-section structure covering lifecycle, nodal points, stakeholders, challenges, package of practices, technology interventions, capacity building, enterprise opportunities, the SELCO opportunity and income impact.
What is going wrong#
The research itself is the bottleneck. Each chain means working through Government of India ministry reports, NABARD and SFAC publications, ICAR material, commodity board data from bodies such as the Spices Board and the Central Silk Board, and state department figures — then extracting comparable information into an identical structure.
SELCO’s quality bar is what makes this hard rather than merely long. Every figure, cost, yield, price and claim must carry a named source with a link or full citation. Unsourced numbers are rejected. Where data does not exist, the gap must be stated plainly rather than filled with a plausible estimate — a draft with an invented number is returned.
Why it is not fixed yet#
Doing 32 of these by hand is roughly 32 times the work of doing one, because nothing carries over. And the two rules that matter most — every figure sourced, every gap declared — are exactly the two things that break down when a tired person is finishing their fourth deck.
Skills and Expertise#
- Branch — CS, IT. Open to any branch for the research side, since the subject matter spans agriculture, animal husbandry, textiles and small business.
- Technical skills — Information retrieval, document parsing and extraction, citation tracking, working with language models under a factual-accuracy constraint, automated document generation.
- Other skills — Desk research at a professional standard, judging source credibility, writing to a fixed structure, working to a review cycle with external deadlines.