For employers

Hire this AI Trainer

Sign in or create an account to invite AI Trainers to your job.

LoginCreate account
Invite to Job
Timhub Consultants

Timhub Consultants

Agency
KENYA flag
Nairobi, Kenya
$10.00/hrIntermediate55+ISO 27001

Key Skills

Software

Data Annotation TechData Annotation Tech
Deep SystemsDeep Systems
Google Cloud Vertex AIGoogle Cloud Vertex AI
HiveMindHiveMind
HumanaticHumanatic
LabelboxLabelbox
LabelImgLabelImg
LionbridgeLionbridge
OpenCV AI Kit (OAK)OpenCV AI Kit (OAK)
RemotasksRemotasks
SamaSama
SuperAnnotateSuperAnnotate
Other
TelusTelus
Scale AIScale AI

Top Subject Matter

No subject matter listed

Top Data Types

Computer Code ProgrammingComputer Code Programming
DocumentDocument
Medical DicomMedical Dicom

Top Label Types

Bounding Box
Computer Programming Coding
Evaluation Rating
Prompt Response Writing SFT
RLHF

Company Overview

Founded by visionary Computer Science students in their university hostels, Timhub Consultants is an emerging BPO powerhouse based in Nairobi, Kenya. We have evolved from a grassroots startup into a specialized data labeling hub, providing high-fidelity ground-truth data for global AI leaders. Our Workforce We operate with a highly optimized team of 55 professionals, maintaining an elite QA-to-annotator ratio to ensure technical precision: 35 First Pass Annotators: Dedicated to accurate, high-speed labeling. 15 QA Specialists: A heavy oversight layer for multi-stage verification. 5 Administrators: Managing operations, security, and growth. Elite Client Portfolio Our technical expertise has earned the trust of industry giants: Autonomous Driving: Data annotation for Cruise. Agri-Tech: Precision plant science labeling for Syngenta. Smart Capture: Supporting Scandit’s advanced scanning tech. Mobility AI: Refining machine learning models for Uber. The Timhub Advantage Rooted in Tech: Being built by CS students gives us a deep technical understanding of the algorithms our data supports. Quality-Heavy Structure: Our 1:2.3 QA ratio ensures near-perfect accuracy that mass-market BPOs cannot match. Secure & Scalable: Based in Nairobi’s "Silicon Savannah," we combine student innovation with professional data integrity and strict security protocols. Mission: To fuel the future of AI by bridging the gap between academic excellence and global industry needs.

IntermediateEnglish

Security

Security Overview

Remote Safety & Security Measures Secure Access (Zero Trust Architecture): Mandatory VPNs: All remote sessions must pass through an encrypted Virtual Private Network (VPN) tunnel, ensuring data in transit is invisible to hackers. MFA Deployment: Multi-Factor Authentication is required for every login attempt, ensuring that even if a password is lost, the account remains secure. Endpoint Security: Device Lockdown: We utilize Endpoint Management software to prevent the use of USB drives, external storage, or screen-recording tools on work devices. Clean Browser Environments: Remote annotators work within sandboxed browser environments that prevent data from being cached or saved locally on personal machines. Real-Time Monitoring & Accountability: Activity Verification: Our admins use non-intrusive monitoring tools to track active vs. idle time, ensuring that the work for clients like Uber or Cruise is being performed in a focused environment. Randomized Quality Check-ins: Our 15 QA specialists perform "live audits" on remote work, jumping into active sessions to verify that security protocols are being followed. Privacy-First Workspace Requirements: Vetted Home Environments: Remote employees must certify a private, distraction-free workspace. Zero-Device Policy: The "No-Phone" rule remains in effect even at home; employees are trained to keep personal recording devices away from their workspace during active sessions.

Security Credentials

ISO 27001

Labeling Experience

Medical Coding and Transcription

OtherTextTranscription
1. Transcription & Coding Medical Transcription: Converting dictations and surgical reports into structured EHR text. Speech-to-Text Editing: Correcting AI-generated drafts for clinical terminology errors. ICD-10/11 & CPT Coding: Assigning alphanumeric codes to diagnoses and medical procedures. Chart Abstracting: Extracting key data points from patient files for billing and research. 2. Clinical Quality Assurance Coding Audits: Verifying code accuracy to prevent claim denials or "upcoding." Safety Verification: Cross-checking drug dosages and medical names against the audio source. Compliance Filtering: Redacting PII/PHI to maintain HIPAA and NDPA privacy standards. EHR Structural Review: Ensuring final reports meet specific hospital system formatting.

1. Transcription & Coding Medical Transcription: Converting dictations and surgical reports into structured EHR text. Speech-to-Text Editing: Correcting AI-generated drafts for clinical terminology errors. ICD-10/11 & CPT Coding: Assigning alphanumeric codes to diagnoses and medical procedures. Chart Abstracting: Extracting key data points from patient files for billing and research. 2. Clinical Quality Assurance Coding Audits: Verifying code accuracy to prevent claim denials or "upcoding." Safety Verification: Cross-checking drug dosages and medical names against the audio source. Compliance Filtering: Redacting PII/PHI to maintain HIPAA and NDPA privacy standards. EHR Structural Review: Ensuring final reports meet specific hospital system formatting.

2025
Telus

SFT

TelusTextText GenerationPrompt Response Writing SFT
Supervised Fine-Tuning (SFT): Drafting "Gold Standard" prompt-and-response pairs to teach models specific tones, formats, and reasoning patterns. RLHF (Reinforcement Learning from Human Feedback): Ranking multiple model-generated responses based on helpfulness, honesty, and safety to refine reward models. Red Teaming & Safety Testing: Deliberately attempting to trigger "hallucinations" or biased/harmful outputs to help Telus build robust ethical guardrails. Complex Multi-turn Dialogue: Writing and evaluating long-form conversations to ensure the model maintains context and "memory" over several exchanges. Fact-Verification (Grounding): Fact-checking model claims against verified external sources to reduce misinformation in generated text. Creative & Technical Writing: Generating specialized content across domains like STEM, finance, and legal to enhance the model's professional utility. Sentiment & Tone Alignment: Labeling text for nuanced emotions to ensure the LLM responds with

Supervised Fine-Tuning (SFT): Drafting "Gold Standard" prompt-and-response pairs to teach models specific tones, formats, and reasoning patterns. RLHF (Reinforcement Learning from Human Feedback): Ranking multiple model-generated responses based on helpfulness, honesty, and safety to refine reward models. Red Teaming & Safety Testing: Deliberately attempting to trigger "hallucinations" or biased/harmful outputs to help Telus build robust ethical guardrails. Complex Multi-turn Dialogue: Writing and evaluating long-form conversations to ensure the model maintains context and "memory" over several exchanges. Fact-Verification (Grounding): Fact-checking model claims against verified external sources to reduce misinformation in generated text. Creative & Technical Writing: Generating specialized content across domains like STEM, finance, and legal to enhance the model's professional utility. Sentiment & Tone Alignment: Labeling text for nuanced emotions to ensure the LLM responds with

2024 - 2025
Scale AI

Evaluation/ Rating

Scale AIComputer Code ProgrammingEvaluation Rating
Pairwise Coding Preference (Side-by-Side): Comparing two model-generated code snippets for the same prompt and ranking them on a 7-point Likert scale based on efficiency and logic.Functional Correctness Testing: Manually executing model-generated code in isolated environments (IDE/Terminal) to verify it compiles and passes all test cases.Prompt Adherence Audit: Evaluating whether the model strictly followed constraints, such as using a specific library (e.g., NumPy, React), a certain time complexity ($O(n \log n)$), or a specific design pattern.Debugging & Bug Identification: Analyzing model-generated code to find "silent bugs," security vulnerabilities, or logical fallacies that standard automated linters might miss.Code Optimization Rating: Scoring models on their ability to refactor "working" code into more memory-efficient or performant versions.Multi-Turn Coding Dialogue: Engaging in back-and-forth interactions with the LLM to refine a complex codebase, testing if the model retain

Pairwise Coding Preference (Side-by-Side): Comparing two model-generated code snippets for the same prompt and ranking them on a 7-point Likert scale based on efficiency and logic.Functional Correctness Testing: Manually executing model-generated code in isolated environments (IDE/Terminal) to verify it compiles and passes all test cases.Prompt Adherence Audit: Evaluating whether the model strictly followed constraints, such as using a specific library (e.g., NumPy, React), a certain time complexity ($O(n \log n)$), or a specific design pattern.Debugging & Bug Identification: Analyzing model-generated code to find "silent bugs," security vulnerabilities, or logical fallacies that standard automated linters might miss.Code Optimization Rating: Scoring models on their ability to refactor "working" code into more memory-efficient or performant versions.Multi-Turn Coding Dialogue: Engaging in back-and-forth interactions with the LLM to refine a complex codebase, testing if the model retain

2023 - 2025

Data Labelling

Other3D SensorBounding BoxSegmentation
3D LiDAR Cuboid Annotation: Boxing objects in 3D point clouds for depth and orientation. Semantic Segmentation: Pixel-level labeling of drivable lanes, sidewalks, and vegetation. Temporal Object Tracking: Frame-by-frame ID linking for trajectory prediction. Edge Case Labeling: Identifying rare objects (e.g., unicycles, construction barriers). Attribute Tagging: Identifying brake lights, turn signals, and flashing sirens. Traffic Light State Verification: Classifying active bulb colors and directional arrows. Lidar-Camera Fusion: Cross-referencing 2D images with 3D sensors for ground-truth accuracy. Motion Prediction Mapping: Annotating pedestrian intent and direction of travel.

3D LiDAR Cuboid Annotation: Boxing objects in 3D point clouds for depth and orientation. Semantic Segmentation: Pixel-level labeling of drivable lanes, sidewalks, and vegetation. Temporal Object Tracking: Frame-by-frame ID linking for trajectory prediction. Edge Case Labeling: Identifying rare objects (e.g., unicycles, construction barriers). Attribute Tagging: Identifying brake lights, turn signals, and flashing sirens. Traffic Light State Verification: Classifying active bulb colors and directional arrows. Lidar-Camera Fusion: Cross-referencing 2D images with 3D sensors for ground-truth accuracy. Motion Prediction Mapping: Annotating pedestrian intent and direction of travel.

2023 - 2023