Design, technology and AI jobs, with client work from week one.
We are an independent design, technology and AI company with offices in India and Canada. Join one team that designs brands, builds websites, apps and AI tools, and runs marketing. AI does the repetitive work. A named person approves what goes live.
This is the first month we plan for every new hire. Each week gives you one reason to join and one thing to show.
Week 1
Release one small change
You join your lead on a client project. Your first change goes live or to the client to check.
Why it mattersWork that goes live
One change · checked by your lead
Week 2
Build an AI agent
In your AI workspace, you build an agent: AI that does one routine task. You test it before anyone uses it.
Why it mattersAI as a tool you control
Agent + 20 test cases · pass rate logged
Week 3
Join another team's review
You sit in on another team's weekly review. An engineer might join a brand critique, for example.
Why it mattersLearn from other teams
Review notes · one question you raised
Week 4
Own a decision, in writing
You make one decision on your part of the project. You log the options, the trade-off and what would change your mind.
Why it mattersSeniority without the politics
Decision-log entry · credited to you
Week 4, up close
A sample decision-log entry.
We log every decision that changes scope, quality or risk, with the reasons and who made it. Reviews start from the log.
DECISION LOG · Client appD-0142
Decision
Agent drafts welcome-screen text. A person approves each market.
Options
A: write by hand · B: agent drafts, person approves · C: agent publishes
Chose
B. C failed 3 of 20 test cases on regulated claims.
Revisit if
C passes all 20 test cases for 2 releases.
Owner · New hire, week 4Approved by lead
A sample decision-log entry. A new hire has an agent draft welcome-screen text, and a person approves each market. Letting the agent publish failed 3 of 20 test cases on regulated claims. The entry also says when to look again.
How we work
Offices in India and Canada, and one way of working across them.
Work from our New Delhi or Ludhiana office. In roles marked remote, work from anywhere in India. The same rules apply everywhere.
Designers pick 6 versions, rewrite 2 by hand and send the flagged claims to legal.
Decide · Your company
approved
6
rewritten
2
approver
Named, logged
6 approved · 2 rewritten by hand
Step 4, Decide, done by person: 6 approved · 2 rewritten by hand.
05 / 05System
The decision goes in the log
The log shows who approved what, which checks ran and which AI model drafted each version.
Record · Your company
entry
D-0142
linked
B-2291
retained
With the project
Decision D-0142 · linked to B-2291
Step 5, Record, done by system: Decision D-0142 · linked to B-2291.
The standards
Six standards behind every hire, review and promotion.
Here is what each one looks like, and where we check it when hiring.
01
Show the work
Evidence over adjectives.
Looks like
A link, a number or a before-and-after
Your own part on a team project, named
What you would do differently now
Does not look like
"Passionate about innovation"
A list of tools without what you made with them
Where we check itApplication read: we open your links before your CV.
02
Write it down
If it changed the work, it is in the log.
Looks like
Decisions with options, trade-offs and a name
Plans a stranger could act on
Short, plain sentences
Does not look like
Decisions made in a hallway and lost
40-slide decks where one page would do
Where we check itWork sample: you add a short note explaining your choices.
03
Check the AI
AI output is a draft until someone has tested it.
Looks like
Testing before trusting, with results kept
Knowing when the AI is guessing
A named approver for anything a client sees
Does not look like
Pasting AI output into client work
Refusing the tools on principle
Where we check itWork sample: use AI if you like, and tell us how you checked it.
04
Say the bad news early
An early warning beats a late surprise.
Looks like
Raising a risk the day you see it
Telling a client the full cost
Owning a mistake in the review
Does not look like
Saying all is well when it is not
Blaming people afterwards
Where we check itConversation: we ask about a decision you got wrong.
05
Finish the details
The last 10 per cent is the job.
Looks like
Rare cases, empty screens, small phones
Proofreading your own work
Taking critique without taking it personally
Does not look like
"Good enough for now" that becomes forever
Polish that hides a weak idea
Where we check itPortfolio or code review: we look at the details you chose to fix.
06
Build for everyone
Accessible, inclusive and private by default.
Looks like
Disability access from the start
Research with people unlike you
Collecting only the data a feature needs
Does not look like
Access checks left to the last week
Designing for the team's own devices
Where we check itDesign review: we ask who your work leaves out.
01
Show the work
Evidence over adjectives.
Looks like
A link, a number or a before-and-after
Your own part on a team project, named
What you would do differently now
Does not look like
"Passionate about innovation"
A list of tools without what you made with them
Where we check itApplication read: we open your links before your CV.
02
Write it down
If it changed the work, it is in the log.
Looks like
Decisions with options, trade-offs and a name
Plans a stranger could act on
Short, plain sentences
Does not look like
Decisions made in a hallway and lost
40-slide decks where one page would do
Where we check itWork sample: you add a short note explaining your choices.
03
Check the AI
AI output is a draft until someone has tested it.
Looks like
Testing before trusting, with results kept
Knowing when the AI is guessing
A named approver for anything a client sees
Does not look like
Pasting AI output into client work
Refusing the tools on principle
Where we check itWork sample: use AI if you like, and tell us how you checked it.
04
Say the bad news early
An early warning beats a late surprise.
Looks like
Raising a risk the day you see it
Telling a client the full cost
Owning a mistake in the review
Does not look like
Saying all is well when it is not
Blaming people afterwards
Where we check itConversation: we ask about a decision you got wrong.
05
Finish the details
The last 10 per cent is the job.
Looks like
Rare cases, empty screens, small phones
Proofreading your own work
Taking critique without taking it personally
Does not look like
"Good enough for now" that becomes forever
Polish that hides a weak idea
Where we check itPortfolio or code review: we look at the details you chose to fix.
06
Build for everyone
Accessible, inclusive and private by default.
Looks like
Disability access from the start
Research with people unlike you
Collecting only the data a feature needs
Does not look like
Access checks left to the last week
Designing for the team's own devices
Where we check itDesign review: we ask who your work leaves out.
Life here
Critique on Tuesdays, demos on Fridays and deep work in between.
Our weekly routine keeps every team working as one, in the office or on calls.
Fig. 01Tuesday critique: paper first, then pixels.
Fig. 02A flow worked out at the board before anyone builds it.
Fig. 03Friday demo: what went live this week.
Fig. 04One-to-ones every 2 weeks. You write the notes.
Fig. 05Wednesday mornings: no meetings before 14:00.
MonTueWedThuFri09:0010:0011:0012:0013:0014:0015:0016:0017:00Core hoursPlans, in writing10:00–10:45Craft critique15:00–16:30Deep work09:30–14:00No meetings · agents run the checksCross-team review16:00–17:00Demo and log16:30–17:30
A typical week
IST · UTC+5:30
Mon10:00
Plans, in writing Today
Each team posts its plan for the week: goals, risks and what it needs from others.
Tue15:00
Craft critique Today
Work in progress goes on the wall, one team at a time. We critique the work, never the person.
Wed09:30
Deep work Today
No meetings before 14:00, so mornings are free for focused work.
Thu16:00
Cross-team review Today
A strategist joins AI testing. A designer joins a campaign review.
Fri16:30
Demo and log Today
What went live, what we learned and what went in the decision log.
Once a quarter, office and remote colleagues spend a week together in one office.
A calendar of the working week, Monday to Friday, 09:00 to 18:00 India time. Core hours, 11:00 to 16:00, are shaded and the five fixed moments of the week are marked; the same five are listed beside it.
What you get
Benefits and an AI workspace, with rules for using it.
This is a draft of what we plan to offer. Your offer letter sets out the final benefits for your role.
Offer letter · Schedule 2
Benefits
Draft · to be confirmed
2.1Health and family
a)Medical insurance
For you, your partner, your children and your parents.
b)Parental leave
Paid leave for every new parent, and a phased return.
c)Mental health
Confidential counselling sessions, paid by us.
2.2Time
a)Paid leave
Earned leave, sick leave and the public holidays of your state.
b)Winter break
Our offices close between Christmas and New Year.
c)Flexible days
Plan your day around the shared core hours.
2.3Growth
a)Learning budget
An annual budget for courses, books and conferences.
b)Growth reviews
Twice a year, against the published standards and levels.
c)Internal moves
Change team or office after a year, with support.
2.4Setup
a)Your own machine
A laptop chosen for your work, replaced every 3 years.
b)Home office
An allowance for a desk, chair and internet that work.
c)Team weeks
Travel and stay paid for the quarterly team week.
Routes and levels
Four ways to join and six levels to grow through.
There is a way in, whether this is your first job or a new field. We write down each level and how to reach the next.
010 open now
Internships
Students and recent graduates
Client work from week one, a named mentor and a written learning plan. We offer strong interns a place on the early-careers track.
Build AI agents and retrieval pipelines for client products, and show they work with evaluation sets, safety checks, approval steps and production monitoring.
Location
New Delhi
Pattern
Hybrid
Level
Mid–Senior
Type
Full-time
New Delhi · Hybrid · Mid–Senior · Closes in 11 days
No. We hire from product companies, in-house teams, research and teaching, as well as agencies. We look for proof of good work, wherever you did it.
02Does software screen the applications?
No. Someone in the team reads everything you send. A person decides whether you go forward. Software only books interviews and searches our own records. You may use AI in the work sample. Tell us how you checked its output.
03Can I work remotely?
Yes, in roles marked Remote: from anywhere in India, with a team week in an office each quarter. In Hybrid roles, you come in on the days your team agrees. We cannot hire outside India for now.
04Is the salary band on each role confirmed?
Yes, once hiring confirms it. Until then, the page marks it as an estimate. You hear the final band in your first conversation, before any task.
05Is the work sample paid?
Yes. Work samples are short, paid and based on the job. You keep what you make. Each role page shows its task.
06How are notice periods handled?
We plan around them. A long notice period is never a reason to turn you down. If buying out your notice suits both sides, we talk about it openly.
07Can I apply for more than one role?
Apply for the closest one and mention the other in your answers. If you would suit another team better, we pass your application on.
08Do you help with relocation?
For office roles, we discuss relocation support case by case. What we agree goes in your offer letter.
09What happens to my data?
We use it only to consider you for roles here. Only the people running your process see it. We delete it six months after the process ends. If you tick "keep on file", we keep it for twelve months to tell you about other roles. The full candidate privacy statement is on the application form.
10Can I ask for an adjustment?
Yes. Ask on the form or by email. You might want extra time, written questions in advance, captions or another format. Asking never counts against you. You do not have to say why.
Next step
If no role fits, tell us what you would build.
Someone in the team you choose reads it like any other application, and replies either way.