Find Work

We need experienced playtesters for our roguelike, which is currently too hard in the first ten minutes and too easy after thirty. We want structured written feedback: where you died and why it felt fair or unfair, which builds felt dominant, where you got bored. Not bug reports — we have QA for that. You should have put real hours into games in this genre. Someone who can say "this feels like Hades but the dodge has no invincibility frames and that is why it reads badly" is exactly who we want. About 8 hours a week.

New office, 2,400 sq ft, warm shell. About 30 seats, two meeting rooms, a phone booth or two and a pantry. We need space planning, material and furniture specification, and photorealistic renders good enough to get our board to sign off. Acoustics matter — our last office was an open plan echo chamber and it genuinely affected how people worked. Please tell us how you would handle that here. Deliverables as CAD drawings a contractor can build from, plus renders.

Setting up a private limited company for a new arm of the business. Two directors, both resident. Need the full sequence handled: name approval, incorporation, PAN and TAN, GST registration, and the opening compliance calendar so we know what is due and when. We have done this once before and lost six weeks to a name rejection and a bounced filing. We are paying for someone who has done it enough times to avoid that.

Eighty apparel SKUs, each needing four angles on white plus one styled shot. Colour accuracy is the thing we care about most. Our return rate on colour mismatch is 8% and a chunk of that is photography, not the product. We want a calibrated workflow and are happy to pay for the extra time it takes. Studio in Mumbai preferred so we can send samples without shipping delays. Retouching included, delivered on our naming convention. Please send apparel work specifically — flat lay and on-model both.

We lend to small businesses and our operations were built for 200 accounts. We are at 600 and everything that was manageable is now painful — collections follow-up, document verification, exception handling. We want someone who has scaled a lending book to sit with us for a few weeks, find what breaks first, and tell us what to fix in what order. Not a strategy deck. We want to be told which three things to change this quarter and why those three. Hourly, roughly 8-10 hours a week.

We are building a 6,000 sq ft cold storage unit near Nashik for fresh produce. We need the layout, chamber sizing across two temperature zones, refrigeration and airflow specification, and a loading dock arrangement that does not create a warm-air bridge every time a truck arrives. Experience with perishables specifically. Cold chain for produce is not the same problem as frozen goods, and we have had one consultant already tell us it was. Drawings a contractor can tender against.

A day of shooting across two of our studios in Pune — classes in progress, equipment details, trainers, and a set of portraits. We need images that look like our actual studio on an actual Tuesday, not a stock-photo gym. Real members, real sweat, good light. Deliverable is around 80 edited images with usage rights for web and paid social. Photographer must be Pune-based or willing to travel at their own cost within the quoted fee.

Our two directors spend an hour a day each on scheduling and triage that someone else could do better. Work is calendar management across timezones, first-pass inbox triage against rules we will agree, meeting notes, and chasing people who owe us documents. We need reliability more than speed, and someone comfortable saying "I did not understand this one, flagging it" rather than guessing on a client email. About 20 hours a week, overlapping IST business hours.

Around 120 tickets a day across chat and email — order status, returns, sizing questions. We will provide the macros and the returns policy. What we cannot provide is judgement about when a customer needs a real answer instead of a template, and that is what we are hiring for. English and Hindi both needed in writing. Freshdesk experience helpful, not essential. Six days a week, 9am to 6pm shift, with one weekday off.

We price packages the way we always have: cost plus a margin, adjusted when a competitor undercuts us. It is almost certainly leaving money on the table in some segments and losing us bookings in others. We want an analysis of our booking data by segment and season, a view of what the market will bear, and a pricing framework we can apply rather than a set of prices that goes stale in a quarter. Three years of booking data available.

We want to offer a nutrition track alongside our classes — a structured 12-week programme our members can follow, built around Indian home cooking rather than chicken and broccoli. Must work for vegetarians and for people eating what the family is eating, because that is most of our members and it is why generic plans fail here. Deliverables: the programme structure, weekly meal frameworks, a shopping approach, and coach-facing notes so our trainers can support it. Qualified nutritionist or dietitian only — we will ask for registration details.

Analytics currently runs against a read replica of production, which is slow for us and occasionally unpleasant for production. We want a proper warehouse: ingestion from Postgres plus two SaaS sources, dbt models with tests, orchestration in Airflow, and documented lineage. We care that the models are testable and that a failed load is loud rather than silent. We have been bitten by a pipeline that half-succeeded for a week. About 40 source tables. Please tell us how you would structure staging and marts before you tell us which tool you like.

Spend sits in Meta and Google, bookings sit in our own database, and nobody can say what a booking costs us by channel without a manual exercise. We want that joined in BigQuery and reported in Looker Studio: spend, bookings, cost per booking and a simple attribution view by channel and campaign. We know last-click is imperfect. We want it stated clearly on the dashboard rather than quietly assumed.

We throw away 12% of fresh stock and we know that number only in aggregate. We cannot say which SKUs, which stores, which days. We want the analysis and a recurring report: wastage by SKU, store, day of week and season, with the biggest recoverable losses called out. The useful output is not a chart. It is a short list of the ten changes that would recover the most, in order.

120,000 customers, three years of order history, and a marketing team that currently segments by "bought in the last 90 days" and nothing else. We want RFM plus behavioural clustering, with segments that are actually actionable — a cluster we cannot write a different email to is not useful to us. Deliverable includes the segment definitions written in plain language for the marketing team, and the code so we can re-run it monthly.

Four years of lending data, roughly 18,000 loans, with repayment outcomes. We want a scorecard we can defend to a regulator and explain to a borrower — which rules out anything we cannot interpret. A gradient boosted model that scores better and cannot be explained is not a better model for us. We need the usual rigour: out-of-time validation, stability over cohorts, and clear documentation of every variable and why it is in there. Please describe a scorecard you have built and how it performed after deployment, not just in validation.

We have Google Analytics and no idea what people actually do inside our product. We want a tracking plan designed before anything is instrumented — event names, properties, naming conventions, and who owns them — then implemented in PostHog. We have seen what happens without that: 400 events, half of them duplicates with different capitalisation, and nobody trusts any of them. We would rather have 30 events we believe.

We receive about 300 documents a day by email — purchase orders, invoices, delivery notes, complaints — and a person sorts them into folders. We want automatic classification and routing, with a confidence threshold below which it goes to a human rather than guessing. Mixed quality: some are clean PDFs, some are photographs of paper taken at an angle. Handling the bad ones well is the actual job. Please tell us how you would measure this in production, not just on a test split.

About 2,800 members across four studios, three years of attendance and billing data. We want to understand who leaves and when, and whether attendance in the first month predicts it as strongly as we suspect. We are not looking for a churn prediction API. We want the analysis and a clear answer to one question: what should we do in a member's first 30 days that measurably keeps them longer.

We review flagged transactions manually against rules written two years ago. Our false positive rate is high enough that the team has started rubber-stamping. We want anomaly detection that reduces the review queue without missing the cases that matter. Precision and recall trade off here and we want to make that trade deliberately, with numbers. We can share a labelled set of confirmed fraud, though it is small — how you handle the class imbalance is a big part of what we are evaluating.