Machine Learning (ML)Location: Bangalore
Experience: 3-5 years
Type: Full-time | On-site
Start: Immediate
Why this role exists
Most companies are using LLMs.
Very few are building an advantage from them.
Right now, LLM cost is our largest margin constraint, and model behavior is still too generic to be defensible.
This role exists to:
- Turn LLM usage into a cost-efficient system
- Build compounding intelligence across accounts
- Create a differentiated analysis layer that competitors can’t replicate
What you’ll do
You will not just build models.
You will own the intelligence and cost structure of the platform.
1. Drive down LLM cost dramatically
- Reduce cost per interaction from ₹40 → ₹2 within 6 months
- Implement:
- Model tiering (right model for the right task)
- Caching strategies (semantic + response caching)
- Batching and async processing
- PTU / reserved capacity optimization
- Ensure performance does not degrade while reducing cost
2. Optimize infrastructure spend
- Reduce cloud spend from ₹20L/month → ₹4L/month
- Work across infrastructure layers (Azure / compute / inference)
- Balance:
- Latency
- Cost
- Throughput
- Treat infra as a first-class optimization problem
3. Build the fine-tuning and learning pipeline
- Design systems where:
- Every interaction improves future performance
- Build pipelines for:
- Fine-tuning
- Feedback loops
- Continuous model improvement
- Ensure the 5th customer deployment is structurally better than the 1st
4. Create a differentiated intelligence layer
- Build analysis systems that:
- Extract signals from interactions
- Improve decision-making
- Drive outcome improvements
- Move beyond responses → insight + action
5. Enable new AI-native product categories
- Identify opportunities where:
- AI enables workflows that were not previously possible
- Build foundational ML capabilities to unlock those categories
- Focus on creation, not just efficiency
6. Commoditize LLM usage internally
- Abstract complexity of LLM usage from product teams
- Build internal systems where:
- Cost is predictable
- Performance is consistent
- Make LLM usage a reliable utility layer
What success looks like
- Cost per interaction drops to ₹2 or lower
- Infrastructure spend reduces 5x without performance loss
- Model performance improves with every deployment
- Platform develops a clear intelligence advantage
- New AI-native capabilities become possible due to your systems
Who you are
- You have 3-5 years of experience in ML / applied AI / systems engineering
- You have worked with:
- LLMs
- Inference optimization
- Production ML systems
- You think in:
- Systems
- Trade-offs (cost vs latency vs quality)
- You care about real-world impact, not just model metrics
What will make you stand out
- Experience with:
- LLM optimization (prompting, fine-tuning, distillation)
- Distributed systems or infra-level optimizations
- High-scale inference systems
- Built systems that:
- Reduced cost significantly
- Improved performance over time
- Strong understanding of:
- Caching strategies
- Model routing
- Evaluation frameworks
Why join
- You will directly impact company margins and scalability
- Your work defines whether we have a defensible ML advantage
- You will build systems that move from:
- Generic AI usage → compounding intelligence
What this role is not
- Not research-only
- Not experimentation without production impact
- Not isolated from product and business outcomes
What this role is
- A builder of ML systems at scale
- A driver of cost and performance optimization
- A creator of long-term competitive advantage
One question to self-evaluate
Can you build ML systems that get cheaper, smarter, and more valuable with every interaction?
Shubham Vishwakarma
Full Stack Developer - Averlon
About Agentic Universe
About
Company social profiles
Similar jobs
Machine Learning (ML)Request for Proposal (RFP): AI Receptionist for Helvetica Incoming Broker Calls
Company Overview
Helvetica Group is a direct commercial real estate lender and investment bank, specializing in providing alternative financing solutions for complex real estate transactions. We focus on fast, innovative, and common-sense underwriting to serve brokers, investors, and borrowers nationwide.
Website: helveticagroup.com
Project Overview
We are seeking an experienced AI Engineer to design and implement an AI Receptionist using Retell AI and Make (formerly Integromat). This AI Receptionist will handle incoming broker calls, collect loan request details, integrate with Salesforce, and automate follow-up communications.
Scope of Work
The solution should:
- Receive Incoming Calls
- Integrate with our phone system and Retell AI to answer broker calls professionally.
- Detect caller intent and initiate a structured loan intake conversation.
- Prompt Loan Request Details
- Use natural language conversation to gather key loan scenario data:
- Borrower and broker information
- Property type, location, value
- Loan amount requested, purpose, and timeline
- Any special circumstances (e.g., foreclosure bailout, BK buyout, etc.)
- Data Handling & Automation
- Send collected call details to Salesforce as a new Lead.
- Email a summary of the loan request to our broker group.
- Send an automated confirmation email to the caller.
- Send an SMS confirmation text to the caller with a thank-you and next steps.
- Integration Requirements
- Use Retell AI for call handling and voice interactions.
- Use Make for workflow automation (Salesforce, email, SMS).
- Ensure all data flows securely and complies with applicable regulations.
Deliverables
- Fully functional AI Receptionist integrated with our systems
- Retell AI call flow design and scripts tailored to Helvetica's lending guidelines
- Make (Integromat) scenarios for:
- Salesforce lead creation
- Email notifications (internal and external)
- SMS confirmation to caller
- Documentation for setup, maintenance, and future scaling
- Optional: Dashboard for call analytics and performance monitoring
Skills Required
- Experience with Retell AI (or similar conversational AI voice platforms)
- Expertise in Make (Integromat) or similar workflow automation tools
- Strong knowledge of Salesforce API integrations
- Experience with Twilio or other SMS/email APIs is a plus
- Understanding of secure data handling and compliance (preferred)
Proposal Requirements
Please include in your proposal:
- Relevant experience with Retell AI, Make, and Salesforce integrations
- Portfolio of similar AI automation projects (voice + CRM)
- Estimated timeline for project completion
- Proposed budget and pricing structure
- Any additional recommendations to enhance this solution
Timeline
- Proposal Submission Deadline: [Insert Date]
- Project Kickoff: [Insert Date]
- Expected Completion: Within 4-6 weeks from kickoff
Budget
We are open to proposals with fixed price or hourly rates. Please provide a clear breakdown of your pricing.
How to Apply
Submit your proposal directly through Upwork, including all requested details.
Would you also like me to include:
- Sample call flow conversation script (to attract engineers who understand the flow)?
- Technical architecture diagram (for clarity on integrations)?
- I can draft both to include in your posting. Shall I proceed?
Machine Learning (ML)- Model Development and Deployment:
- Design, develop, and implement machine learning models for various applications (e.g., classification, regression, natural language processing, computer vision).
- Deploy and maintain machine learning models in production environments.
- Optimize model performance and efficiency through feature engineering, hyperparameter tuning, and model selection.
- Build and maintain scalable machine learning pipelines.
- Data Engineering and Management:
- Work with large datasets, including data cleaning, preprocessing, and feature extraction.
- Develop and maintain data pipelines for data ingestion, transformation, and storage.
- Ensure data quality and consistency.
FURIOUS FOX is looking for Embedded Developers with strong coding skills in C & C++ as well as experience with Embedded Linux.
Experience : (Minimum 7-10 yrs)
• Experienced in edge processing for connected building / industrial / consumer
appliances / automotive ECU
• Have a good understanding of IoT platforms and architecture
• Deep experience in operating systems eg: Linux, freeRTOS / kernel development/device drivers.
/ sensor drivers
• Have experience with various low-level communication protocols, memory devices, messaging
framework etc.
• Have a deep understanding of design principles, design patterns, container preparations
• Have developed hardware, OS abstraction layers, and sensor handlers services to manage various BSP, os standards
• Have experience with Python edge packages.
• Have a good understanding about IoT databases for edge computing
• Good understanding of connectivity application protocols and connectivity SDK for Wi-Fi and BT / BLE
• Experienced in arm architecture, peripheral devices and hardware board configurations
• Able to set up debuggers, configure build environments, and compilers and optimize code and performance.
Skills / Tools:
• Expert at object-oriented programming
• Modular programming
• C / C++ / JavaScript / Python
• Eclipse framework
• Target deployment techniques
• IoT framework
• Test framework
Highlights :
• Having AI / ML knowledge in applications
• Have worked on wireless protocols
• Ethernet / Wi-Fi / Bluetooth / BLE
• Highly exploratory attitude
• willing to venture in and learn new
technologies.
• Have done passionate projects based on self-interest.
PythonSolinas Integrity (www.solinas.in) is a leading water & sanitation robotics start-up founded by IIT Madras Alumni & professors to develop cutting edge solutions to solve the problems in water pipelines and sewer lines\septic tanks, thereby improving the lives of millions of people. Our core values start with trust, and respect for everyone and along with strong collaboration and communication. We believe in giving agency to our teammates and strongly pushing them towards developing a growth mindset.
Duties and Responsibilities:
- To develop and improve signal processing algorithms for analysis of acoustic signals with up-to-date knowledge on processing methods.
- Understand key acoustic algorithm functions, develop efficient code, verify performance and functionality.
- Exposure to all phases of software development life cycle (concept, design, implementation, test, and production).
- Propose innovations to improve performance, quality, etc.
- Work with peers to develop excellent, structured code, well-optimized and easily maintainable.
Basic Qualifications:
● Experience programming in either Python, C++, or MATLAB
● MS/PhD degree in Electrical/Electronics Engineering/ Signal processing
● At least 1 year of signal processing or related area
● Good analytical and problem-solving skills
● Good knowledge of signal processing techniques, basic knowledge of ML algorithms and good visualisation skills.
Follow CutshortShubham Vishwakarma
Full Stack Developer - Averlon