Manoj Kumar
Data Annotator | AI/ML Data Operations
Hire this person
Send a job offer directly to this candidate
About
Detail-oriented Data Annotation Specialist with 1 year of professional experience in 2D image annotation and 3D LiDAR video annotation, plus 5 months of freelance data annotation experience from May–September 2026. Skilled in image classification, bounding box and polygon annotation, segmentation, object detection and tracking, point-cloud labelling, LLM response evaluation, content categorization, multilingual data projects, and QA/QC validation. Experienced in PII/document validation and AI/ML training-data workflows with strong accuracy, guideline compliance, and attention to edge cases.
Experience
Freelance Data Annotator | RAI/LLM Content & PII/Privacy Annotation - Freelance
(2026-05 - 2026-09)
- Performed dual-track annotation for a Responsible AI content-safety and privacy-protection initiative, spanning harmful-content labelling and PII detection across multilingual, multi-domain text.
- Classified prompt/completion pairs across six harm categories (Misconduct, Violence, Hate, Stereotype, Insults, Sexual) following a structured, version-controlled SOP.
- Scored each item on intent (harmful/harmless/unspecified), intensity (low/medium/high/hard-to-decide), a 0–5 severity scale, and a 0–4 confidence rating, applying nuanced rules for multi-label harms, profanity vs. insult distinctions, and edge cases (incomplete text, multilingual content, text-processing requests).
- Delivered labels in a standardized output format to maintain consistency across a distributed annotation team.
- Annotated PII with character-exact span offsets across healthcare records, benign look-alike data (for false-positive-rate reduction), long-context documents requiring coreference resolution, and structured data (JSON/XML/HTML).
- Applied 15+ canonical PII entity types with locale-specific variants (US, UK, Canada, India, international) and delivered output in structured JSONL format meeting >95% annotation-accuracy QA standards.
- Incorporated iterative client feedback to improve span accuracy, adopt newly introduced entity tags, and increase realism of document/conversation samples.
Data Annotator - Adept Talent Acquisition Pvt Ltd
(2025-05 - 2026-05)
- Executed 2D and 3D annotation tasks for autonomous driving, logistics, and mapping datasets.
- Created accurate labels for vehicles, pedestrians, traffic signs, speed limits, stop boards, curbs, medians, and road edges.
- Performed sensor fusion alignment between camera and LiDAR data for precise localization.
- Maintained high accuracy and consistency while meeting daily productivity targets.
- Supported QA teams by reviewing annotations and correcting errors.
- Daily annotated and validated 1000+ 3D LiDAR point cloud frames and objects (vehicles, pedestrians, cyclists, road infrastructure) using 3D bounding boxes and cuboids with a 98.5% QA accuracy rate.
- Placed and adjusted accurate 3D bounding cuboids for dynamic road users (vehicles, pedestrians) with strict adherence to project ground-truth rules.
Education
Bachelor of Technology - B. Tech - Annamacharya Institute of Technology & Sciences, Tirupati (AITS-T) (2016)