Tech Talk Interviews

Can you explain what algorithmic trading is and how it differs from traditional trading methods?

Bijit: Algorithmic trading is a preset configurable computer program which basically, based on different variables like time, volume, and market data, executes different sets of trade programs based on the time dimension, based on the program dimension, based on the cost dimension, and based on how you want to program it accordingly.

Traditionally, trading is mostly human led, which is driven by human intuition, human signals, human evaluation, and analysis. That's the biggest difference in algorithmic trading versus human led trade.

How does risk management factor into algorithmic trading systems?

Bijit: Risk and overall analysis is a part of our overall flow, how you want to make a decision or how you want to go from analysis to capture the trading systems. It's not an afterthought. These trades happen in micro milliseconds. The error exposure is very less, and the room for error is also very non negotiable in that nature.

We want to make sure that speed is a part of our pipeline, risk is a part of our pipeline. The explainability, where we want to trace from human decisioning to data decisioning and completely make it data driven, is completely baked into the pipeline from starting with the data to the trade getting captured, the trade getting settled as an overall life cycle of the trade.

What recent technological advancements in AI are you most excited about for trading?

Bijit: I'm excited about the agentic approach in the complete life cycle of trade management, starting with the agent doing the pre trading risk analysis and the agent augmented human led trading systems where the agent is extracting a lot of risk analytics data, giving a lot of recommendations based on the market dimensions.

Then there's the post trade systems where the agent is helping in reconciliations and giving a shift from T plus 2 to T plus 0 and giving the liquidity optimizations. That's the exciting piece for the overall market dynamics.

As algorithmic trading becomes more autonomous, what are the primary ethical challenges you're seeing?

Bijit: The primary challenge as this algorithmic trading becomes more autonomous is explainability. We need to make sure that every decision, every analysis is traced through from the data source. There's a lot of synthetic non generative data coming into the picture. How do we want to clean that generative data when 80% of the decisions come from that? That's a very big exciting piece for us.

But if an agent is making those levels of decisions, that would be hard for humans to analyze. Previously I was talking about how the room for error is very less. We need to make sure that we have a complete trace, we have an auditable trail as a part of the decision, and most importantly we have an explainability piece for the human as domain experts.

What advice would you give to financial institutions looking to adopt AI and ML systems in their trading operations?

Bijit: My primary advice would be keeping compliance from day zero. Compliance by design, compliance by code, so that you have the complete understanding where you really have to dynamically orchestrate your compliance, not as an afterthought but as the dynamics of data is changing. The compliance also needs to evolve during that state.

The second piece is explainability. Be more responsible about what decisions humans need to make versus what agents need to make. I think that separation is very equally important until we have complete confidence into how the agent needs to operate in that life cycle and making sure that the responsible and ethical nature of autonomous AI agents is having complete oversight as a part of the evaluations.

How do you see the role of human traders evolving alongside these AI systems?

Bijit: The human element remains critical, especially in terms of domain expertise and oversight. What we're seeing is not a replacement but an augmentation. The AI agents handle the speed, the data processing, the pattern recognition at scales that humans simply cannot match. But humans provide the strategic thinking, the ethical guardrails, and the ability to understand context that goes beyond pure data.

The key is defining clear boundaries. What decisions should remain human led? What can be safely delegated to agents? This isn't a static answer. It evolves as our confidence in the systems grows and as the technology matures.

What's your outlook for the next few years in this space?

Bijit: I'm super excited about this domain. I think 2026, 2027, and the next three years will be completely driven by agents as we see them becoming more mainstream. I firmly believe that trading systems are also going to evolve with the latest nature of agentic approaches.

What should organizations be preparing for now?

Bijit: Organizations need to invest in three areas simultaneously. First, the technical infrastructure that can support these high speed, data intensive operations. Second, the governance frameworks that ensure compliance and explainability are built in from the start, not bolted on later. And third, the talent that understands both the domain of trading and the capabilities and limitations of AI systems.

The institutions that get ahead will be the ones that view this as a holistic transformation, not just a technology upgrade. It's about reimagining workflows, redefining roles, and rebuilding trust mechanisms for an AI augmented environment.

We're at an inflection point where the technology is mature enough to handle real world trading complexity, but we're still early enough that the organizations making smart investments now will have a significant competitive advantage. The speed of execution, the quality of risk management, the efficiency of capital deployment, all of these are being redefined by AI.

But success won't come from just deploying the most advanced algorithms. It will come from thoughtfully integrating these systems with human expertise, maintaining rigorous compliance standards, and never losing sight of the fact that behind every trade are real stakeholders whose trust we must earn and maintain.

Conclusion:

Ghosh's perspective highlights a critical balance: embracing the transformative power of AI in trading while maintaining the explainability, compliance, and human oversight that financial markets demand. As agentic systems become mainstream, the institutions that succeed will be those that build responsible AI into their foundation rather than treating it as an afterthought.