JD #2403
Applications open
Full-Time Job
Project Cursa - Robot Manipulation Video Annotator
Welo Data
About the role
Welo Data is an emerging AIβdriven data solutions company that focuses on building highβquality training datasets for robotics, computer vision and natural language processing. Founded by a team of robotics engineers and data scientists, the firm has quickly become a trusted partner for global robotics manufacturers looking to accelerate the learning curve of their autonomous systems. With a culture that blends technical rigor with creative problemβsolving, Welo Data encourages its employees to experiment, iterate and deliver data that powers the next generation of intelligent machines.
The role of AI Video Annotator is a cornerstone of Welo Dataβs Project Cursa, a flagship initiative aimed at creating richly annotated video corpora for robot manipulation tasks. As an annotator, you will be the eyes and voice of the dataset, translating raw robot footage into structured, naturalβlanguage descriptions that feed directly into machineβlearning pipelines. This position is fully remote, allowing you to work from anywhere in India, the Philippines, Mexico or Egypt, while collaborating with a globally distributed team of data curators, robotics engineers and AI researchers.
Key Responsibilities
1. Watch short robot manipulation videos captured from three synchronized camera angles (overhead and two wristβmounted views).
2. Segment each video into atomic motions, skills/subβtasks, and overall tasks, ensuring every second is labeled at two or more levels.
3. Write clear, concise, naturalβlanguage descriptions for each segment, adhering strictly to a detailed style guide.
4. Accurately capture both successful actions and failures (e.g., dropped objects, missed grasps).
5. Crossβreference all camera angles to verify object identities, contact points and spatial relationships.
6. Apply consistent vocabulary for actions, spatial terms (left/right, above/below) and object parts (handle, edge, lid).
7. Participate in periodic calibration sessions to align labeling quality with the team and client expectations.
8. Provide feedback on the style guide and suggest improvements based on edgeβcase scenarios.
9. Maintain a high level of productivity while working independently, meeting daily and weekly annotation targets.
10. Use annotation tools such as Label Studio to upload, edit and export labeled data.
Tech Stack & Tools: Label Studio (or equivalent), Google Drive for video sharing, Excel/Google Sheets for tracking, basic familiarity with robotics terminology.
Growth Path: Starting as a freelance annotator, high performers can progress to Senior Annotator, Quality Lead, or even transition into Data Engineering or Robotics QA roles within Welo Dataβs expanding product teams.
Why Join Welo Data? You will contribute directly to cuttingβedge robotics research, gain exposure to AI data pipelines, and develop a niche skill set in video annotation that is in high demand across the AI industry. The flexible remote setup, competitive hourly pay, and a supportive community of data experts make this an ideal entryβpoint for fresh graduates eager to build a career in AIβenabled data engineering.
Eligibility & process
Who can apply
Graduates (any discipline) with a minimum of 60% aggregate in the final year. Candidates from the 2024, 2025 or 2026 batch are preferred. No active backlogs at the time of joining. Strong command of written English is mandatory.
Skills & stack
Selection process
- Round 1: Online aptitude / logical reasoning test
- Round 2: Annotation task and styleβguide assessment
- Round 3: HR interview (fit, availability and compensation discussion)
Prep kit
19 resourcesStudy material 3
Aptitude Practice Questions
Helps you prepare for the initial logical reasoning test commonly used by Welo Data to assess analytical ability.
Company Interview Preparation Guide
Provides insights into typical interview formats and questions asked by techβdata companies like Welo Data.
General Interview Preparation Resources
Covers behavioral and situational interview tips useful for the HR round at Welo Data.
Company-specific guides 4
APTITUDE & LOGIC
Aptitude Practice Questions & Online Mock Tests
Curated logical, quantitative, and verbal reasoning problems to sharpen reasoning skills required for the initial screening test.
COMPANY GUIDE
Welo Data Interview Preparation Corner
Comprehensive guide covering typical interview formats, common questions, and preparation tips for Welo Data and off-campus tech roles.
MOCK TESTS & PAPERS
Comprehensive Interview Prep Resources & Syllabus
Collection of previous year questions, company-specific test patterns, and interview experiences for technical and HR rounds.
CODING PRACTICE
Algorithm and Data Structure Problem Set
Practice problems to improve coding proficiency and algorithm problem-solving speed for technical rounds.
Likely coding questions 4
Round-wise strategy 4
Round 1: Online Assessment (OA)
Aptitude, Quantitative Logic & 2 Coding Problems
Focus on accuracy and speed. Practice arrays, strings, and standard arithmetic puzzles.
Round 2: Technical Interview 1
Core Tech Stack (Excellent Written English, Video Annotation, Structured Labeling, Spatial Description, Basic Robotics Terminology, SelfβDiscipline, Ability To Follow Style Guides, Next.js) & Live Code Tracing
Explain your thought process aloud. Analyze time & space complexities before coding.
Round 3: System Design & Problem Solving
Database Schemas, APIs & Architecture Basics
Clarify edge cases, diagram schemas cleanly, and discuss scalability trade-offs.
Round 4: HR & Cultural Fit Discussion
Welo Data Core Values, Learning Agility & Offer Terms
Demonstrate passion, strong communication, and readiness for full-time collaboration.
Technical questions 5
What is the difference between WHERE and HAVING clauses in SQL? Answer
Model answer: WHERE filters rows before any groupings are applied, while HAVING filters aggregated groups after GROUP BY has executed.
Explain SQL Window functions: ROW_NUMBER(), RANK(), and DENSE_RANK(). Answer
Model answer: ROW_NUMBER() assigns unique sequential integers. RANK() assigns identical ranks to ties and skips ranks. DENSE_RANK() assigns identical ranks to ties without skipping rank numbers.
How do you handle NULL and missing values during data cleaning in Python/Pandas? Answer
Model answer: Use .isna().sum() to identify missing values. Impute with mean/median using .fillna() or remove with .dropna(subset=[...]) depending on variance impact.
What is the difference between Star Schema and Snowflake Schema in Data Warehousing? Answer
Model answer: Star Schema has denormalized dimension tables directly connected to the central Fact table. Snowflake Schema normalizes dimension tables into sub-dimensions to minimize redundancy.
How do you calculate MoM (Month-over-Month) growth in SQL? Answer
Model answer: Use LAG(revenue, 1) OVER (ORDER BY month) to fetch the previous month's revenue and compute (revenue - prev_revenue) / prev_revenue * 100.
Behavioural & HR questions 3
Why do you want to join Welo Data as a Project Cursa - Robot Manipulation Video Annotator?
Highlight Welo Data's market reputation, recent tech innovations, and how your skills in Excellent Written English directly solve their team's objectives.
Highlight Welo Data's market reputation, recent tech innovations, and how your skills in Excellent Written English directly solve their team's objectives.
Describe a challenging bug or academic project roadblock and how you resolved it.
Use the STAR method: Situation (project context), Task (what needed solving), Action (specific tools/logic applied), Result (quantifiable positive outcome).
Use the STAR method: Situation (project context), Task (what needed solving), Action (specific tools/logic applied), Result (quantifiable positive outcome).
How do you handle strict deadlines or sudden scope changes?
Explain your prioritization strategy, proactive communication with mentors/peers, and agile mindset.
Explain your prioritization strategy, proactive communication with mentors/peers, and agile mindset.