Sr. ML Engineer - Consultant

About Ushak Kaal

Ushak Kaal Communications Limited is a technology company that builds and runs applied AI systems for businesses. We work with clients to put AI agents into real customer conversations and make them measurably better over time.

The Role

We are hiring one hands-on senior engineer to build a shared brain for our AI agents across texting, calling and email. It should remember past lessons, use our team's feedback, and steadily improve how conversations turn into bookings.

The Problem Today

When an agent makes a mistake, we capture the conversation and edit its instructions by hand. Those fixes don't last or carry across channels. We need every correction to become a lasting lesson that guides future texts, calls and emails.

Customer accounts stay in the existing CRM. This project reads from it and adds a memory of lessons learned on top.

What You Will Build

The system should work like coaching a new employee: show it what went wrong, explain what to do instead, save the lesson, and check that it handles similar situations better next time.

• Capture every conversation. This covers text messages, email threads, call transcripts and relevant call audio, plus confirmed bookings and every point where a person had to step in.
• Set goals and rules per channel. Define what good looks like for text, calls and email, including correct pricing, valid bookings and appropriate call handling.
• Review conversations automatically. Check them against those rules, flag likely problems, and give our team a simple way to mark what was right or wrong.
• Store approved corrections as lessons. Each lesson holds the situation, the better response or action, the reason, and which channels it applies to.
• Apply lessons in live conversations. Use relevant lessons in future texts, calls and emails, share them across channels where they fit, keep channel-specific guidance separate, and always use current business and pricing information.
• Test before release. Run changes against past mistakes and new situations in all three channels, compare results with the current agent, and keep a one-step way to roll back.

How Success Is Measured

Success means more confirmed bookings and fewer repeated mistakes across all three channels, measured against today's starting results.

• Booking rate: the share of conversations that end in a confirmed booking.
• Repeated mistakes: how often an error that already has a lesson happens again.
• Human intervention: routine conversations run with zero hand-holding, and exceptions hand off cleanly to a person.
• Accuracy: correct pricing, valid bookings and appropriate call handling.

Who We Are Looking For

We want a senior AI or machine learning engineer who has shipped AI agents used by real customers and can write the software behind them. You will own this system and work with the existing engineering team to connect it to our live agents.

Must Have

• Hands-on experience connecting AI models to texting, calling and email systems, including OpenAI APIs and live voice agents.
• Experience storing lessons or feedback and testing improvements against human judgement.
• Strong Python or TypeScript, plus solid database skills.
• Ability to build a simple internal screen where the team reviews mistakes and enters corrections.
• Clear communication, with real examples of improving a live AI agent using feedback and measurable results.

Nice to Have

• Experience with booking, scheduling or sales-conversion flows.
• Familiarity with CRM integrations and consumer-facing service businesses.